1

The Anisotropy of Personal Space

Jhilmil Nigam

Applied Psychology, SDSOS; NMIMS

Environmental Psychology

Ms. Anushka Merlyn Arakal

October 20, 2022

 2

Abstract

Personal space violations are associated with discomfort. However, the precise

function that links the magnitude of discomfort to interpersonal distance is yet to be

determined. In this study, we investigate whether interpersonal distance and discomfort are

isotropic concerning uncomfortably far or close distances. We also extend previous findings

regarding intrusions into personal space and distance maintenance outside of personal space.

We presented 15 interpersonal distances ranging from 40 to 250 cm to subjects and collected

verbal and joystick-based ratings of discomfort. Whereas discomfort increased immediately

upon entering personal space, the gradient was less steep for distances that exceeded the

personal space comfort region. As a result, personal space is anisotropic in terms of

discomfort.

The Anisotropy of Personal Space

When a stranger approaches us, we begin to feel uneasy and intruded upon. Our

perception of an inappropriately large or short distance from another person can be

interpreted as a personal space requirement accompanied by a psychological distance.

Personal space was first studied in animals to see how it extends and shapes distance

behavior. Animals in captivity, for example, claim a smaller territory and flight zone than

wild animals. Sommer pioneered human proxemic research. When interacting with others in

a hospital, he noticed that schizophrenic patients claimed a larger portion of space to

themselves than non-schizophrenic patients. Hall expanded on the concept of interaction

distances by proposing four distinct intimate spaces (0-45 cm), personal space (45-120 cm),

social space (120-365 cm), and public space is defined by their radius, which is primarily

determined by the appropriateness of potentially available sensory perceptions (365–762 cm).

 3

These ranges have been replicated across a wide range of nationalities and cultures, as well as

in various measures of interpersonal distance.

Leslie Hayduk provides the most prominent definition of personal space: "... we can

define personal space as the area individual humans actively maintain around themselves into

which others cannot intrude without causing discomfort." This definition organized proxemic

research and improved conceptualization, personal space measurement, and correlation

identification.
 4

Method

24 subjects were recruited at the University of Mainz aged from 18 to 28 years (M =

21.66, SD = 6.92, 6 male), with an average body height of 170.96 cm (SD = 7.25 cm) (Robin

Welsch et al, 2019). Before testing, a consensual survey was held in Helsinki following the

declaration of Helsinki, and filled out a demographic questionnaire.

Design and Stimuli

Subjects were placed at 15 frontal IPDs to a confederate ranging from 40 cm to 250

cm in 15 cm steps, which corresponds to Williams' mean minimum and maximum distance

for conversation [11]. These distances were marked—but not labeled—on the floor with tape.

On any given trial, both the subject and the confederate were placed on a random pairing of

these marks that were aligned with their body center. The body's center was estimated to be in

the middle of the foot, as indicated by dots on the shoes. Throughout the experiment, subjects

and the confederate were instructed to look each other in the eyes. Both of the confederates in

this study were young females. One of the confederates was 165 cm tall. Subjects were

placed at 15 frontal IPDs to a confederate ranging from 40 cm to 250 cm in 15 cm steps,

which corresponds to Williams' necessarily imply minimum and maximum distance for

conversation. These distances were marked—but not labeled—on the floor with tape. On any

given trial, both the subject and the confederate were placed on a random pairing of these

marks that were aligned with their body center. The body's center was estimated to be in the

middle of the foot, as indicated by dots on the shoes. Throughout the experiment, subjects

and the confederate were instructed to look each other in the eyes. Both of the confederates in

this study were young females. One of the confederates was 165 cm tall. The other was 167

cm tall with brown hair and blond hair. The two collaborators alternated between sessions to
 5

account for potential confounding variables such as fatigue, poor concentration, and so on.

Both confederates were dressed in white shirts and blue jeans.

Procedure

We standardized the social situation for all testing blocks to reduce situational effects

on IPD [3]. Subjects were asked to imagine themselves in an open space in an unfamiliar city,

asking a stranger for directions. In a fixed-distance task, subjects were placed at 15 different

IPDs and asked to rate subjective discomfort verbally on a rating scale ranging from -100

(maximum discomfort, too close) to 0 (ideal distance) to +100. (maximum discomfort, too

far). In Block 1, the experimenter directed the subject while the confederate remained

stationary. In Block 2, the subject remained stationary while the confederate switched

positions between trials. During the positioning, the subjects were blindfolded. The blindfold

was removed after positioning and the results were subjective.

Block 3 was similar to Block 2, but subjects rated discomfort by positioning a joystick.

Because the confederate couldn't see the exact tilt of the joystick, this was done to control

social desirability. When IPD was deemed too close, subjects were instructed to tilt the

joystick away from themselves as a function of discomfort or to tilt the joystick towards

themselves when the distance was deemed insufficient. All possible Blocks 1, 2, and 3 orders

were used and counterbalanced between subjects. The order of distances within each block

was randomized. To estimate the preferred IPD, subjects completed two repetitions of an

active and a passive stop-distance task. In the active stop-distance task, the subject

approached the confederate until he or she reached a comfortable IPD. In the passive

stop-distance task, the confederate approached the subject slowly until the subject signaled

the confederate to stop. Subjects were given the option of fine-tuning this distance by

instructing the confederate to move forward or backward. The preferred IPD was measured
 6

on the floor with a tape measure and recorded as the distance between the subject's and the

confederate's body centers. Within the sample, the order of the passive and active

stop-distance tasks was counterbalanced. Individual 60-minute sessions were used to test the

subjects. There were no time constraints in any of the trials. The subjects were thanked and

debriefed following the procedure. In this study, we report on all measures and scale

manipulations. We did not exclude any of the experimental trials from data analysis, and we

did not increase the sample size after data analysis.

Statistical Analysis. We chose a Bayesian approach to data analysis to allow us to test

the Null-hypothesis. The Bayes Factor (BF) is used for statistical inference and is calculated

in R using the BayesFactor-package. Given the observed data, the BF quantifies the relative

likelihood of the Null-model versus the alternative model. We either give the likelihood of the

Null model in comparison to the alternative model (BF01) or the reverse fraction (BF10). In a

prior-sensitivity analysis, we compared various weakly informative priors. The priors used in

this study had no effect on statistical inference because the data obtained clearly outweighed

the priors when computing the Bayes Factors. As a result, in t-tests, we used the

BayesFactor-default package's priors, regressions, and variance analyses We present 95%

confidence interval median estimates of parameters with high-density intervals from the

posterior distribution. We used BRMS, a wrapper for the STAN-sampler for R, to calculate a

Bayesian linear mixed model (BLMM) to model the relationship between IPD and

discomfort. On all beta-coefficients, we used normally distributed priors (M = 0, SD = 1),

with Cholesky priors on the residual correlation ( = 1) and a t-distributed prior to allow for

thicker tails (df = 3, M = 0, SD = 10) on the centred intercept, variance parameters, and

sigma. These priors are only very weakly informative and serve primarily to regularise the

posterior distributions. We ran four Hamilton-Monte-Carlo chains, each with 10,000

 7

iterations and 20% warm-up samples. Divergent transitions were identified using trace plots

of Markov-chain Monte-Carlo permutations. All Rubin-Gelman statistics were significantly

lower than 1.1.

Results. First, we will consider relative reliability, which refers to the stability of rank

order and differences within the sample, and then we will consider absolute reliability, which

refers to the absolute deviation of sequential measurements.

Test-retest reliability, and thus relative reliability, as measured across two repetitions, was

high in both the passive, =.85 [.70;.93], BF10 > 100, and active conditions of the

stop-distance tasks, =.94 [.87;.97], BF10 > 100. In terms of absolute reliability, Figure 2

shows that the individual mean IPD of both trials, as well as the differences between

repetitions 1 and 2, were largely independent in both tasks (panels A and B). This implies that

within-subject variation was unrelated to variation between subjects, and thus within-subject

variation was most likely unsystematic.

An intrusion into PS of 15 cm or more beyond the comfort point causes a sharp increase in

discomfort. Movement in the opposite direction away from the other person causes a similar
 8

but shallower increase in discomfort. As a result, the reaction to intrusion and extrusion is

anisotropic. Equal distances from PS's border result in unequal increases in discomfort. The

gradient of intrusion is steeper than that of extrusion. Let us illustrate this point with the

example of a conversation between two people, A and B. If person A reduces IPD toward

person B, the likelihood of B taking a corrective action away from person A should increase

immediately.

In other words, we anticipate hysteresis in the following sense. In both active and passive

approaches, we have always used an approach scenario in which the initial IPD was greater

than the ideal IPD. We would expect a slightly larger preferred IPD in the latter case if the

stop-distance task was started from a position well within the intrusion zone and once from a

position well within the extrusion zone. It is worth noting that the anisotropy of PS holds true

for intrusion/extrusion but not for active/passive approach. They found no differences

between active and passive approaches, contrary to Iachini et al. This could be due to

subjects' habituation to our stimulus. While our colleague completed all experimental trials,

their stimuli changed at random throughout the experiment. Thus, the effects of perceived

dominance or potential fear of the approaching target, which may be especially noticeable in

active approach, may have faded in our experiment. This could also explain their

experiments' relatively large (i.e. more conservative) judgments of preferred IPD. We can

generate qualified hypotheses about the effects of a given person variable or a given

environment variable within this field-theoretical framework. For example, based on what is

known about psychopathy, it might make sense to hypothesise that the equilibrium-point is

unaltered in psychopathic subjects, but the gradient is much shallower on the intrusion side

than it is in less psychopathic subjects when confronted with social threat. External factors,

such as the crowdedness of the space, on the other hand, could simply shift the equilibrium
 9

point without affecting the gradient's steepness. In a crowded market, for example, the

equilibrium-point should be closer to the person. Future research should refine this

field-theoretical model and test the novel predictions it allows for in terms of discomfort and

IPD.
 10

References

Welsch, R. (2019, June 4). The anisotropy of personal space. PLOS ONE. Retrieved October

20, 2022, from

https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0217587

Hayduk, L. A. (1983). Personal space: Where we now stand. Psychological Bulletin, 94(2),

293–335. https://doi.org/10.1037/0033-2909.94.2.293