Does who I am and what I feel determine what I see (or say)? A meta-analytic systematic review exploring the influence of real and perceived bodily state on spatial perception of the external environment

Distance estimation (n = 5)

All experiments evaluated verbal distance estimation data. Pooling of two experiments (Baati et al., 2020; Baati et al., 2015) that compared distance perception pre- vs post-exercise using a sprint cycling task (10 × 6 s maximal cycling) found no effect of exercise at any of the three distances measured, for either the normal (Fig. 3A) or the sleep deprived conditions (Fig. 3B). Within the same studies, there was an effect of fatigue due to sleep deprivation only at the 35 m distance (Fig. 3C), where participants perceived the distance as closer when undergoing sleep deprivation compared to a normal night sleep. A final study (Hunt, Hunt & Park, 2017) evaluated the effect of fatigue (stepper machine 90 s) vs non-fatigue, finding that the fatigued group estimated a single target distance of 92.5 m as being farther than the non-fatigued group (Hedge’s g: 0.56, 95% CI [−0.95 to −0.16]).

Two experiments used comparable distance measures, but in athletic populations (Asaf, Santillán & Barraza, 2015; Jarraya et al., 2013). Asaf, Santillán & Barraza (2015) showing no difference (Hedge’s g range: −0.13 to 0.09) in distance estimations based on their fatigue manipulation (treadmill running 2 min), whereas Jarraya et al. (2013) found very large effect sizes following cycling for 10 mins (Hedge’s g range: 8.32 to 10.08), indicating participants estimated distances as closer in the post-exercise (fatigued) condition compared to the pre-exercise condition. Pooling of the studies’ similar distances (12 m and 11 m) was non-significant, with high heterogeneity (I² = 96%). Large differences in effect sizes at a similar distance (g = −0.14 vs 9.92), suggest that between study differences are unlikely driven by differences in the spatial perception task; but rather, may lie in the fatigue manipulation. Sensitivity analyses evaluating the influence of varying imputed correlation coefficients showed similar results (Supplemental File 5, Tables S1–S4).

Hill steepness estimation (n = 3)

Three experiments (Bhalla & Proffitt, 1999; Proffitt et al., 1995; Taylor-Covill & Eves, 2013) evaluated the effect of fatigue (all running to exhaustion/maximal fitness tests) on steepness estimation of hills (n = 2) or stairs (n = 1) and were able to be pooled. For verbal measures of steepness (Fig. 3D), participants estimated hills as steeper after fatiguing exercise for both shallow (5°) and steep hills (15–31°). For haptic measures of steepness (Fig. 3E), participants also estimated shallow hills as steeper after fatiguing exercise, but not steep hills. Sensitivity analyses, as described above, showed similar results (See Supplemental File 5, Tables S5 and S6).

Interim discussion

There was inconsistent and conflicting evidence for an effect of fatigue on spatial perception of distance, although consistent evidence was seen for hill steepness, with fatigue resulting in reports of increased hill steepness.

For distance spatial perception, pooling found no effect of fatigue, as induced by exercise (Fig. 3A), exercise and sleep deprivation (Fig. 3B) or sleep deprivation alone (Fig. 3C). However, our risk of bias appraisal identified potential issues with fatigue paradigms (e.g., insufficient fatigue for perceptual scaling to occur) and/or distance estimation tasks (e.g., distances not sufficiently challenging for the population). For example, in both pooled studies (Baati et al., 2020; Baati et al., 2015) that found null results, a relatively easy sprint cycling task (6 s of sprinting × 10), particularly for a young population (mean = 23 years), was used to induce fatigue. In contrast, Hunt, Hunt & Park (2017) recruited an older population (mean = 54 years), used a challenging exercise task for that population (stepper task), and used longer distances (92.5 m), and did show findings consistent with EoA hypothesis: fatigued participants estimated distance as further than non-fatigued participants. Exercise task difficulty in context of the population may also have relevance to findings (Jarraya et al., 2013) that were contrary to what EoA would predict (fatigued athletes estimated distances as being closer than when they were not fatigued). When a highly trained athletic population undertakes a task that is not sufficiently challenging to induce fatigue, it is possible that this increases activation/arousal and perceived energy/vigour (Thayer, 1978), inducing an opposite perceptual shift. Given that the studies did not assess whether their fatigue manipulations were successful, it is difficult to know if they were adequate to induce perceptual scaling consistent with the EoA. However, these inconsistent findings may also provide evidence against EoA. Indeed, given high risk of bias for participant blinding and given the populations recruited (e.g., often athletes vs non-athletes), social effects on compliance with experimental demands and knowledge effects cannot be ruled out. In particular, athletes (such as the soccer players in both Baati et al. (2015, 2020) experiments), may have substantial knowledge/experience in estimating distances. Thus, we cannot rule out the effect of such knowledge on the results.

The “difficulty” of the spatial perception task may also support more consistent findings seen here for hill steepness estimation. Hills are more difficult to walk than flat surfaces (hills requiring more capacity), thus hill estimation may require less ‘fatigue’ to induce a perceptual shift than flat distance measures. Indeed, pooling of verbal estimations of hill steepness found that people who were fatigued typically estimated hills as being steeper than control groups/conditions (consistent with EoA), although findings were less clear when using haptic measures. Further work is needed to clarify such findings, particularly given that many pooled studies were small, conducted by the same research groups, and the actual fatigue level induced was rarely evaluated.

Distance estimation (n = 3)

Two studies comparing walking distance estimation in people with and without persistent pain (Tabor et al., 2016; Witt et al., 2009) were pooled. There was no significant effect of pain on distance estimation (Fig. 4). High heterogeneity was seen (e.g., 73%, 83%), where Witt et al. (2009) found that people with lower limb pain verbally estimated distances as farther than pain-free controls, and Tabor et al. (2016) found no differences between pain and pain-free groups. The third study (Alaiti et al., 2019) evaluated the influence of shoulder pain on ratio judgements during a pointing task, where participants estimated the distance to various points on a board (see Table 1 for details). There was no significant effect of shoulder pain on distance estimation (g = −0.33, 95% CI [−0.68 to 0.02]), but a trend towards those with shoulder pain estimating distance as farther than those without shoulder pain was present.

Interim discussion

There was limited and inconsistent evidence for an effect of pain on distance perception. The risk of bias assessment identified potential importance of pain by task interactions i.e., that the location of body pain might need to be relevant to the actual performance of the spatial perception task. Participants in Witt et al. (2009) had lower limb pain and showed evidence of perceptual scaling for walking distance estimations. In contrast, Tabor et al. (2016) recruited people with a variety of chronic pain conditions in various body locations, including those whose pain might not interfere with the person’s capacity to traverse the environmental distances presented to them, and had null findings. That the spatial perception task might require specificity to the site of pain (i.e., site of reduced bodily capacity), is supported by results found by Alaiti et al. (2019) of a trend towards pain-induced effects on a distance perception task that was specific to the area of pain.

Interestingly, a secondary analysis within Tabor et al. (2016) showed no distance estimation differences between those that anticipated experiencing pain during walking vs those that did not, although it is likely that this post hoc analysis was underpowered. Alternatively, pain may be a proxy for another bodily state – e.g., pain-related functional limitation that results in reduced cardiovascular fitness – that interacts with pain or solely drives changes to spatial perception. Our risk of bias assessment found that none of the studies in this grouping controlled for any confounding variables. Therefore, future studies evaluating numerous putative contributors to spatial perception are warranted.

Distance estimations (n = 8)

Six experiments evaluated verbal estimations of distance and were sufficiently similar to be pooled (Bian & Andersen, 2013; Costello et al., 2015; Sugovic & Witt, 2013). This meta-analysis revealed that older participants estimated distances as farther than younger participants, at all five distances examined (Fig. 5A). There was also a negative effect of age on distance perception at three out of five distances in the single experiment that used an action-based measure (Bian & Andersen, 2013); see Fig. 5B. An equidistant cone task was nonsignificant at all distances (Norman et al., 2020).

Hill steepness estimation (n = 2)

Two experiments (Bhalla & Proffitt, 1999; Dean et al., 2016) examined the effect of age on measures of hill steepness but were unable to be pooled. They both found a positive relationship when estimating the steepness of moderate hills (e.g., 9 or 10 degrees), indicating that the elderly group estimated the hills as less steep than the young group. In an LMM, Dean et al. (2016) reported a beta coefficient for age of −0.2°/year, indicating that hill steepness estimates decreased by 0.2° every one year of increased age. However, while Bhalla & Proffitt (1999) also found that older participants estimated the hills as less steep than younger participants for shallow hills (see Fig. 5C), the opposite was seen for steeper hills (25 or 29 degrees) where older participants estimated hills as steeper than younger participants.

Interim discussion

While there is consistent evidence that age shifts distance perception in line with the EoA (older participants perceive distances as farther than younger participants), the evidence for hill steepness perception is limited and less consistent. Similar to findings in fatigue studies, a potential task-related influence may be present whereby shallow hills may not be sufficiently “difficult” to induce a challenge to homeostatic state, and thus, show an effect of age, whereas steep hills do show an effect of age. Findings from pooled distance estimations are supportive of an influence of ‘difficulty:’ older participants over-estimate distance to a greater extent than younger participants (12 m; g = −1.48 (95% CI [−2.04 to −0.91])) and this age-induced difference is smaller for shorter distances (4 m; g = −0.75 (95% CI [−1.19 to −0.30])). However, this does not fully explain findings that older individuals perceive shallow hills as less steep than younger individuals.

Potential limitations exist for pooled distance estimations with four of the six experiments from the same paper (although four different participant groups). While this raises the possibility of systematic error biasing the pooled result, the observed effects in the four experiments were smaller than those found in the others. Thus, if a systematic bias was present, it would be in the direction of a null effect, which provides increased confidence that findings of an effect of age on distance perception are robust. Our risk of bias assessment indicates that potential confounders were poorly controlled in this grouping (all but (Dean et al., 2016) scoring high/unclear in this risk of bias domain). Therefore, the observed differences between groups may be due to knowledge effects – where older participants have greater experience in hill and distance estimations than younger participants. Conversely, the between-group study designs employed in this grouping may have reduced response bias, as participants would be less likely to guess the study’s hypothesis/anticipated direction of effect.

Distance/aperture width estimations (n = 5)

Two experiments evaluating the effect of increased hand size (via padded gloves) on a visual matching aperture width measure (Collier, 2018) were pooled, finding no effect (Fig. 6A). Contrasting results were seen when foot size was manipulated in virtual reality (Jun et al., 2015): those in the large foot group estimated gaps as smaller than those in the small foot group. Similarly, across two experiments van der Hoort, Guterstam & Ehrsson (2011) used augmented reality (AR) to perform a “body swap” and found that taking on a small body resulted in over-estimation of distance, compared to a normal size body, using both verbal and blind-walking measures. Further, taking on a large body led to underestimation of distance (Fig. 6B). Sensitivity analyses evaluating the influence of varying imputed correlation coefficients showed similar results (See Supplemental File 5, Table S7).

Hill steepness estimations (n = 2)

Two experiments from one study (Bridgeman & Cooke, 2015) compared hill steepness estimation between conditions of standing on the ground vs standing on a 37 cm box. Pooled estimates found that participants standing on a box estimated the hill as less steep than participants standing on the ground, at 2 and 4 m, but not at 8 or 16 m (see Fig. 6C).

Interim discussion

While evidence for an effect of body size manipulation on spatial perception was inconsistent, this finding likely reflects high heterogeneity in body manipulation methods, including potential use of inadequate manipulation paradigms. Three of the seven experiments found an effect that supports the EoA at every distance evaluated (and for both verbal and action-based measures), whereby increasing body size decreased perceived distance, and decreasing body size increased perceived distance. Of these, two experiments assessed body size via embodiment of AR avatars and confirmed the success of their AR experimental manipulation (participants embodied the large/small bodies). The remaining four experiments had null or mixed results, although lack of manipulation checks limits the ability to interpret their findings. For example, Collier (2018) found that padded gloves (vs no gloves) did not change spatial perception. Without a manipulation check, it is unclear whether use of padded gloves solely influenced perceived hand size, whether it resulted in a feeling of encumbrance, or whether it did not alter bodily perception at all. All studies in this grouping scored high or unclear on participant blinding in the risk of bias assessment. Given that most studies in this grouping used within-group designs, lack of blinding may lead to response bias (i.e., participants guessing the hypothesised direction of the effect after being exposed to both manipulation conditions), reducing confidence in these studies’ results. Finally, limitations in the experiments manipulating eye height (via participants standing on a box) make it difficult to determine whether the results provide evidence against EoA or inconsistent results are due to task constraints. The lack of consistent effect on estimated hill steepness for all distances was interpreted as lack of support for EoA (all points on a hill being equally difficult to climb). However, that repeated verbal estimates of the same, unchanging physical hill were used, suggests possible response bias contribution.

Distance estimations (n = 3)

Two experiments (Phillips et al., 2010; Ries et al., 2008) compared distance estimation (via blind-walking) of people embodying an avatar (vs no avatar) in virtual reality (VR). Pooling found that the avatar group estimated distances as farther than the no-avatar group (Fig. 7A). A third study (Scandola et al., 2019) investigated the influence of wheelchair embodiment in people with spinal cord injury (SCI). There was a negative effect of embodiment on distance perception (via distance on hill): people estimated distances as farther for steep hills when in their own (embodied) wheelchair vs a generic (non-embodied) wheelchair; there was no effect for distances on shallow hills (Fig. 7B).

Hill steepness estimations (n = 1)

Scandola et al. (2019) also evaluated the effect on hill steepness perception of people with SCI and found no difference between using their own (embodied) wheelchair vs a generic (unembodied) wheelchair for visual matching estimation of steepness at any hill angle (Fig. 7C).

Interim discussion

Findings were consistent that embodiment (either of an avatar or of a wheelchair) results in perceptual scaling for distance estimates (makes distances seem farther). A strength of these studies was that embodiment paradigms were experimentally evaluated, allowing confidence that the experimental manipulation or condition evaluated the construct it intended to. Only one study (Scandola et al., 2019) explored the influence of embodiment on hill steepness perception, therefore replication is needed. Two studies in this grouping (Phillips et al., 2010; Ries et al., 2008), were conference abstracts and scored poorly on the risk of bias assessment (both scored high or unclear in all categories except for outcome reporting and missing data), thus reducing confidence in the results.

Distance estimation (n = 5)

These experiments were unable to be pooled due to methodological differences, including type of distance estimation measure (action-based, via blind-walking and bean-bag toss, and visual matching tasks). Four experiments found that participants in the glucose group estimated distances as closer than did those in the control group (placebo glucose; Fig. 8A) (Cole & Balcetis, 2013; Zadra, 2013; Zadra, Weltman & Proffitt, 2016). One experiment had a non-significant result (Cole & Balcetis, 2013), but the results were consistent with the overall direction of effect seen in the other studies.

Hill steepness estimations (n = 5)

All five studies evaluated verbal hill steepness estimation (Durgin et al., 2012; Schnall, Zadra & Proffitt, 2010; Shaffer et al., 2013; Zadra, 2013) and were able to be pooled (Fig. 6A). Ingesting glucose resulted in less steep verbal estimation of hill angle (compared to placebo) for shallow hills, but not steeper hills. There was high heterogeneity in pooled analyses of steep hills (I2 = 81%) with one study (Schnall, Zadra & Proffitt, 2010) finding a large effect in favour of glucose reducing perceived hill steepness, and one with null results (Shaffer et al., 2013).

Interim discussion

There was consistent evidence that participants who ingested glucose estimated distances as shorter and hills as less steep than in control groups/conditions – in line with the EoA. While a non-significant pooled effect was found for the steep hill subgroup, the study that found a small non-significant effect (Shaffer et al., 2013) was assessed to have a higher risk of bias (did not control for confounders, and missing data >15%) than the study that did find a significant effect of glucose (Schnall, Zadra & Proffitt, 2010), thus more weight is placed on the latter’s findings. Furthermore, in all but one study (Cole & Balcetis, 2013), participants ingested food/drink either containing glucose or a sugar free sweetener (placebo) allowing participant blinding. In addition, this work contained the only two experiments (Cole & Balcetis, 2013; Zadra, 2013) where assessors were blinded. Adequate blinding and the use of between-group study designs means that competing theories (experimental demands and response bias) are less likely to have influenced the results, increasing confidence that the findings do reflect an effect of glucose on measures of spatial perception.

Distance estimation (n = 4)

Methodological differences prevented pooling. All effects sizes were negative, but only two significant (Fig. 9A). One study (Sugovic, Turk & Witt, 2016) evaluated the effect of BMI on verbal distance estimation finding significant effects for estimates at 10 m, but not at any other, longer distances, as well as no effect of perceived body size (picture selection task) at any distance. Taylor, Witt & Sugovic (2011) found that traceurs/parkour athletes (i.e., higher fitness/sport-specific training) estimated vertical wall height as significantly lower than untrained controls for a 1.94 m wall but not for higher walls (2.29 m, 3.45 m). Effect sizes were unable to be calculated for two other studies, but one found a non-significant relationship between waist-to-hip ratio and distance perception (Cole, Balcetis & Zhang, 2013), and one found no relationship between self-rated fitness and a distance on hill measure (Shaffer, Greer & Schaffer, 2019).

Hill steepness estimations (n = 7)

Three experiments (all from the same paper) that evaluated the effect of self-rated fitness on verbal (Fig. 9B) and haptic (Fig. 9C) measures of hill steepness were able to be pooled (Krpan & Schnall, 2014) and produced non-significant results. The verbal measure trended towards a positive effect, indicating that people who rated their own fitness as higher, also perceived the hill as steeper; while the haptic measures trended in the opposite direction, towards a negative effect.

The remaining studies used objectively measured weight/fitness and found effects on perceived hill steepness (Fig. 9D). Sugovic (2013) found that hills were estimated as steeper as BMI increased. Similarly, Bhalla & Proffitt (1999) combined BMI and a submaximal fitness test to create a composite “fitness measure”, and found that as fitness increased, estimated hill steepness (verbal) decreased. Haptic responses had no relationship with fitness measurements. In another experiment (Taylor-Covill & Eves, 2016) experimenters used weight coding to group participants into “clearly overweight” and “healthy” groups, and also found that overweight status increased perceived hill steepness estimates. Likewise, Dean et al. (2016) found that people with higher BMI gave higher slant estimations (LMM beta = 0.5). A final experiment (Taylor-Covill & Eves, 2016) also found an association (via LMM) between body composition (measured via dual-energy x-ray absorptiometry [DEXA] at baseline and follow-up; average of 407 days between assessments) and perceived steepness of staircases. At both timepoints BMI and fat mass were positively associated with verbal and visual measures (increases in BMI/fat mass associated with increased estimated steepness), but the model was non-significant for haptic measures of steepness. Longitudinally, when BMI and body fat decreased, so too did verbal estimations of steepness, but not haptic.

Interim discussion

There was no evidence for a relationship between self-rated fitness and spatial perception, but there was consistent evidence for associations with objective fitness measures.

Limitations in the measurement of self-rated fitness likely contribute to the lack of effect. For example, Krpan & Schnall (2014) assessed perceived fitness (self-rated) using a 5-point scale, and had poor distribution of responses (e.g., in experiment 1, 82.7% of participants rated their condition as either good [4] or excellent [5]), and thus, likely has poor sensitivity to differences in perceived fitness. The perceived fitness outcome measures used by Sugovic, Turk & Witt (2016) were similar, with poorly defined constructs (e.g., body type category of Large/Muscular) and limited response options. Unclear validity of the fitness measure cannot be ruled out as a contributor to the lack of relationship between fitness and distance measures.

While more consistent evidence in line with EoA was present for objective measures of fitness, such as BMI, waist-to-hip ratio, and fat mass influencing spatial perception, methods of assessing objective fitness were also problematic. Authors reported measuring “physical abilities” via BMI (Sugovic, Turk & Witt, 2016) and “physiological fitness” via waist-to-hip ratio (Cole & Balcetis, 2013). While these measures hold correlations with measures of fitness (Wier et al., 2006), they have important limitations that may be relevant to EoA. For example, BMI does not account for differences between muscle and fat. Thus, a heavily muscled, but fit individual (e.g., Lebron James, BMI 27.5; overweight category), would have a higher BMI then a sedentary individual, with low muscle mass and high body fat, but we would expect very different results for spatial perception based on the EoA hypothesis (Nevill et al., 2006). One study evaluated “body composition” (Taylor-Covill & Eves, 2016) using DEXA scanning, which does allow for the delineation between muscle and fat. Indeed, this study found that fat mass, rather than fat free mass, was significantly associated with measures of spatial perception both cross-sectionally and longitudinally. Additionally, consistent with other groupings, these studies scored high risk of bias for the blinding domain. Thus, it is difficult to rule out the influence of response bias in these results.

Distance and gap estimation (n = 11)

Pooling found that external loads resulted in participants estimating distances as farther than no load groups/conditions when using verbal (Fig. 10A) and VR based walking distance matching (Fig. 10B) measures, but not blind walking measures (Fig. 10C). An additional study evaluated within-group differences in verbal estimates of distance when wearing a backpack (Vinson et al., 2017), with findings consistent with pooled results (unable to calculate effect size). The remaining experiments were also unable to be pooled (Fig. 10D). Of these, two (Corlett, Byblow & Taylor, 1990; White, 2012) compared medium load (5 kg ankle weight/moderate resistance band) conditions to high load (10 km ankle weight/heavy resistance band) conditions. Both experiments had small, non-significant effects in the direction that the high load group perceived distances as farther during blind walking measures (Corlett, Byblow & Taylor, 1990) and VR based distance matching (White, 2012). Last, one experiment found that participants verbally estimated gap widths to be wider while wearing ankle weights than they did without (Lessard, Linkenauger & Proffitt, 2009). Sensitivity analyses evaluating the influence of varying imputed correlation coefficients showed similar results (Supplemental File 5, Tables S8 and S9).

Hill steepness estimations (n = 1)

A single study evaluated the effect of external loads on verbal estimation of hill steepness (Bhalla & Proffitt, 1999), finding significant effects for a 5° hill but not a 31° hill (Fig. 10D). Opposite effects were seen for haptic measures of hill steepness.

Interim discussion

There was consistent evidence that external loads (vs no load) influence verbal and VR-based measures of spatial perception in line with EoA, but conflicting evidence for studies evaluating varying degrees of loading. Results require cautious interpretation given experimental demands and response bias. When two experiments (Keric & Sebanz, 2021) used identical protocols, but one used a cover story (successfully blinding participants) and one did not (all participants guessed study hypothesis), the effect estimate was heightened in favour of the unblinded group in line with EoA predictions (g = −0.32 vs −0.01 in the blinded group). According to our risk of bias assessment, participants were not blinded in the remaining studies in this grouping. Given that these unblinded studies have similar, large negative effects to the unblinded Keric & Sebanz (2021) experiment, confidence in the overall pooled effect is reduced. Further use of within-group, repeated measure design of some studies (e.g., (Lessard, Linkenauger & Proffitt, 2009)) increase the possibility of response bias driving the effects seen.

Pooled results for VR experiments manipulating external load also showed a significant effect in line with EoA predictions; however, all studies were from the same author (i.e., four comparisons from two experiments), and are from a dissertation (not peer reviewed). Given sample size reductions necessary for pooling, this resulted in small samples (n = 9–17), increasing the risk of spurious results (Higgins et al., 2021). Additionally, assessor participation in measures of spatial perception (e.g., assessor stopped the treadmill during VR-based walking distance matching), paired with assessor non-blinding to test condition/study aim heightens the possibility of measurement bias in this method.

Studies (Corlett, Byblow & Taylor, 1990; White, 2012) investigating varying degrees of external loads (high, low, no load) had non-significant results, but failure to assess participant-specific perceptions of load magnitude hampers clear conclusions. It is unclear whether trends towards increased distance estimation with high vs low loads represent the possibility of a dose-response relationship, perhaps uncaptured if the loads applied were too light to induce a homeostatic threat/influence (e.g., standard high load may not be perceived as high/difficult for some people).

Distance estimations (n = 1)

The high interoceptive accuracy group (two-thirds of participants) perceived the same distance as significantly farther on a hill than on the flat, in line with the EoA (hills require higher energetic cost, therefore the same distance should be perceived as further than on flat), but the low interoceptive group (remaining one-third of participants) did not (Tenhundfeld & Witt, 2015). Between-group differences were significant at only one of the four distances (Hedge’s g at 2 m = 0.25 (95% CI [– 0.35 to 0.85]), 4 m = −0.52 (95% CI −1.12 to 0.09]), 6 m = −0.64 (95% CI [−1.25 to −0.03]), and 8 m = −0.6 (95% CI [−1.21 to 0.01])).

Hill steepness estimations (n = 1)

A significant association was found between interoceptive accuracy and verbal estimations of hill steepness (g = −0.89, 95% CI [−1.36 to −0.41]), indicating that those with higher interoceptive accuracy had lower hill steepness estimations (over-estimated steepness to a lesser degree) (Mouatt et al., 2020, unpublished data).

Interim discussion

Very limited evidence exists that an individual’s ability to perceive their own internal bodily state influences spatial perception. One study (Tenhundfeld & Witt, 2015) had high risk of bias (all domains assessed as high/unclear), and during correspondence, it was reported that a replication study failed to demonstrate an effect (N. Tenhundfeld, 2021, personal communication). Thus, results here should be interpreted with caution.