Exploring the nature of visual fixations on other pedestrians

Abstract

How we look at other people may affect conclusions drawn about the effect of changes in lighting when this task needs to be done after dark. This paper reports further analysis of the distance and duration of fixation on other pedestrians, updating a previous review by considering a greater number of fixations and by examining the influence on these of other characteristics such as the relative direction of travel. This analysis provides further support for a tendency to fixate others at a distance of 15 m and for a duration of 500 ms.

1. Introduction

Pedestrians tend to visually fixate other pedestrians^1,2 and this may be done at least partially to inform the decision as to whether it is safe to approach that person or that they should be avoided.^3–6 After dark, in locations where pedestrians are expected, road lighting should aid this task.

Research has been carried out to investigate how changes in lighting affect the ability to make judgements about other pedestrians. Some studies have used a stop-distance procedure in which the test participant walks towards a target person and stops at the point when the required task (e.g. identification) can be done.^6–12 Further studies have sought judgements when targets are presented at one or more fixed distances.^13–20 The stop-distance procedure reveals the distance at which a task could be completed, but does not reveal whether that distance approximates that at which it is desirable for pedestrians after dark. Experiments seeking judgements at fixed distances may exhibit better ecological validity but only if the distances used in trials include (or bracket) that which approximates pedestrians’ desired distance for evaluating other people.

Caminada and van Bommel⁶ established the basis of current lighting recommendations for pedestrians and within this they adopted the distance estimates of Hall,²¹ i.e. a distance of 3.7 m (12 feet) as the minimum distance at which an alert subject would be able to take evasive or defensive action if threatened, and 10 m, an ideal facial recognition distance being the transition point between the close and far phases of Hall’s public zone. Fotios et al.²² reviewed Hall²¹ and concluded that the evidence was not sufficiently robust and furthermore that Hall had not intended its use for such a purpose. They also found disagreement with other estimates of interpersonal distance²³ and demonstrated range bias in experiments investigating comfortable interpersonal distances.^24,25

Experiments using a stop distance procedure (along with some using fixed distances) require that the test participant continuously fixate the target person. This does not represent typical behaviour. Glances toward other people tend to be short because it is often uncomfortable to look at others for long periods²⁶ and because there may be no need to – information about others such as trustworthiness can be gained from exposures as short as 50 ms.^27,28

As the light level increases (within the range of levels typical of outdoor lighting) then reaction time decreases and visual acuity increases, i.e. objects can be seen more quickly and when they are smaller or further away.²⁹ The accuracy of recognition judgements is affected by distance, becoming more accurate for closer targets.³⁰ Therefore, distance and duration matter to research of lighting for interpersonal judgements because they may affect the lighting conditions found to be optimal for this task. As a first estimate of desirable conditions, Fotios et al.²² drew estimates of distance and duration from eye tracking carried out in a natural outdoor setting.³¹ They found a tendency to fixate upon other people at a distance of 10.3 m and for a duration of 480 ms. When considered alongside other data these estimates were modified to a desire to fixate other people at a distance of 15 m and for a duration of 500 ms. There are two limitations to their analysis of the eye-tracking data: (1) They did not examine all of the available video records, i.e. only two of the four route sections and for the duration estimate this considered only the records for five people; and, (2) They did not record supplementary data such as the group size, gender and relative travel direction of the observed people and hence whether these factors affect the distance and duration of fixation.

This paper reports the findings of a more extensive analysis, made possible by further research funding, extending the previous study by inclusion of all sections of the available video records and by extending the range of data recorded. The key questions targeted in this paper are at what distance and for what duration do people tend to look at other pedestrians, is there a tendency to fixate either the head or body and are these affected by characteristics such as the gender and travel direction of other people?

2. Method

Further analysis was carried out of previously reported work in which mobile eye tracking was used to investigate the gaze behaviour of pedestrians walking outdoors along an urban route of approximately 900 m, in daytime and after dark.^1,31 The 40 pedestrians all followed a near-identical route, completing it in both directions on two separate occasions, resulting in the collection of 80 eye-tracking videos. A number of these sessions provided relatively poor eye-tracking signal due to loss of the pupil image, largely due to individual eye and pupil characteristics and the data being recorded in an outdoor environment with variable weather conditions (see Holmqvist et al.³²). Sessions were excluded from further analysis if the proportion of valid data with acceptable point-of-regard location was below 50%. This resulted in the exclusion of 26 of the 80 videos, the remaining 54 being retained for this analysis.

Each video was coded by a primary coder to record details of every fixation to another pedestrian. Fixation of a pedestrian was defined as the gaze cursor being located on or very near to another pedestrian visible in the video, to allow for the ±1° margin of error present in the eye-tracker (as reported in the manufacturer’s documentation), for at least 3 frames, or 120 ms based on the video’s frame rate of 25 Hz. Table 1 lists the details that were recorded about each fixation. Distances were estimated with reference to the known size of other objects in the field of view, such as vehicles, paving slabs and street benches.

Table 1

Details recorded about pedestrian fixations and reliability test scores between the two coders

Fixation detail	Possible values	Inter-rater reliability test and score
Fixation number (e.g. first, second, third etc. fixation of that pedestrian)	1 to n	N/A
Location	Head, Body, Unknown	Cohen’s Kappa, k = 0.54
Direction of travel	Towards, Away, Unknown	Cohen’s Kappa, k = 0.82
Gender	Male, Female, Unknown	Cohen’s Kappa, k = 0.51
Group	Solitary pedestrian, Pedestrian in group	Cohen’s Kappa, k = 0.85
Fixation duration	In frame numbers, converted to ms	Two-way mixed, agreement ICC = 0.67
Distance	Estimated distance to pedestrian, rounded to nearest metre	Two-way mixed, agreement ICC = 0.69

Classifications for describing Kappa values from Landis and Koch³⁴: <0.00 = poor, 0.0–0.20 = slight, 0.21–0.40 = fair, 0.41–0.60 = moderate, 0.61–0.80 = substantial, 0.81–1.00 = almost perfect. Classifications for describing ICC values from Cicchetti³⁵: <0.40 = poor, 0.40–0.59 = fair, 0.60–0.74 = good, 0.75–1.00 = excellent

To check whether the data recorded by the primary coder were reliable, a second coder blind-recorded the same details for 288 fixations to other pedestrians. Inter-rater reliability analysis to compare the two coders was carried out using Cohen’s Kappa for categorical variables and Intra-Class Correlations (ICC) for continuous variables, as described in Hallgren.³³ The resulting test values and agreement classifications are shown in Table 1. Kappa values ranged between 0.51 and 0.85, which suggest ‘Moderate’ up to ‘Almost Perfect’ agreement between coders according to Landis and Koch.³⁴ ICC values were 0.67 and 0.69, suggesting agreement was ‘good’ according to thresholds set out by Cicchetti.³⁵ Based on the agreement classifications, agreement between the two coders on location and gender are moderate, fixation duration and distance are good and direction of travel and group are almost perfect. Overall, this suggests there was acceptable agreement between the primary and second coders, and data recorded by the primary coder have been used in all subsequent analysis.

Of the recorded measures, distance to a fixated pedestrian is one of the more subjective details estimated.³⁶ While recorded data showed good consistency between different coders, the accuracy to which these reflect the real distance is not revealed. A second validation step was therefore carried out. The actual distances between observer and fixated pedestrian were measured in-situ for 15 situations where positions of observer and fixated pedestrian could be accurately established from the eye-tracking videos and these were compared with estimated distances for those situations by the primary coder. The mean ratio of actual to estimated distances was 1.31 (s.d. = 0.38), with distances being consistently underestimated (mean estimated distance =10.3 m, mean actual distance = 13.9 m, distances were overestimated in only 2 of the 15 situations). This is consistent with previous findings that egocentric distances are underestimated in virtual environments.³⁷ The mean ratio of actual to estimated distances for the 15 situations used in the validation exercise (1.31) was therefore applied to all estimated distances to account for the expected systematic underestimation: this increased the mean estimated distances of these 15 situations to 13.5 m, which is within 3% of the mean actual distance of 13.9 m.

3. Results

Fifty-four eye-tracking videos were included in the analysis, which covered 21 day sessions and 33 after-dark sessions from 37 participants. Within these videos 2496 other pedestrians were fixated, some of whom were fixated more than once, resulting in details being recorded for 5955 separate fixations. Four aspects of fixation were analysed: the number of fixations, the duration of fixations, the distance of first fixation, and whether fixation was on the head or the body. The fixation details were further analysed by the variables included in Table 1 and the gender of the observer.

Analysis of the data was carried out by calculating the mean number of fixations, mean fixation duration and mean fixation distance as recorded within each video, for each pedestrian for their night and day videos separately. These data were found to be normally distributed as assessed by inspection of normality plots, comparison of mean and median, skewness and kurtosis statistics and statistical tests against normal distribution. Parametric statistical tests have therefore been applied as appropriate.

3.1. How frequently do we fixate other pedestrians?

Observers fixated other pedestrians a mean of 2.4 times (s.d. = 0.6); Figure 1 shows the distribution of the mean number of fixations per person. This was significantly influenced by the direction of travel of the fixated pedestrian, with a repeated-measures analysis of variance (ANOVA) suggesting a significant difference between categories (p < 0.001). A post-hoc Tukey HSD test suggested those pedestrians walking away from the observer attracted significantly more fixations (mean = 3.3) compared with those standing still or walking towards the observer (means = 2.1 and 2.0, respectively, p < 0.001 in both cases). This may be because pedestrians walking away from the observer are likely to be in the field of view for a longer period of time, giving the opportunity for more fixations, or because walking away means there is no possibility of eye-contact, this giving a more comfortable situation for looking at other people.^26,38 It may also be because movement away is less familiar than movement towards and requires a longer time to make recognition-type decisions.³⁹ In addition to the higher mean number of fixations on people walking away from the observer, there is also a higher standard deviation: this may be an indication that there is a common motivation for fixating people approaching (considered alongside the lower mean, this might be a common desire to avoid the gaze of others) but a number of possible motivations for fixating people who are stationary or moving away.

Figure 1

Mean number of fixations towards other pedestrians (N = 54). Note: bin labels on the abscissa identify the upper limit of a bin range, e.g. ‘2’ is the range > 1.5 but ≤2.0 fixations

The mean number of fixations towards the body (mean = 1.3) tended to be slightly higher than towards the head (mean = 1.1): this difference is close to being significant (p = 0.06). No other significant effects were found (Table 2). Regarding the gender of the fixated person, while these data did not suggest a significant difference in frequency of fixation toward males or females, the standard deviation is greater for females than for males. This might indicate a common motivation for looking at males (perhaps since considered to be the gender more likely to present a threat) but a number of possible motivations for fixating on females.

Table 2

Mean number and standard deviation of fixations on pedestrians by different variables

Variable		Mean number of fixations	Standard deviation	Significance
Light condition	Day	2.6	0.78	n.s.^a
Light condition	After-dark	2.3	0.51
Gender (observer)	Male	2.4	0.50	n.s.^a
Gender (observer)	Female	2.4	0.76
Gender (fixated pedestrian)	Male	2.5	0.71	n.s.^b
Gender (fixated pedestrian)	Female	2.7	1.30
Location on fixated pedestrian	Head	1.1	0.50	p = 0.06^b
Location on fixated pedestrian	Body	1.3	0.64	p = 0.06^b
Group	Solitary	2.4	0.72	n.s.^b
Group	Group	2.5	0.99
Direction of travel	Towards	2.0	0.53	p < 0.001^c
	Stationary	2.1	1.53
	Away	3.3	1.59

Independent t-test

Dependent t-test

Repeated-measures ANOVA

3.2. How long do we fixate other pedestrians?

The mean duration of each fixation was 475 ms (s.d. = 124 ms); the distribution of mean fixation durations is shown in Figure 2. There was no significant difference between the mean duration of the first fixation of a pedestrian (mean = 477 ms) and any subsequent fixations to that same pedestrian (mean = 492 ms).

Figure 2

Mean duration of fixations towards other pedestrians (N = 54). Note: bin labels on the abscissa identify the upper limit of a bin range, e.g. ‘360’ is the range >240 but ≤360 ms

Viewers tend to concentrate their fixations, including the very first fixation in a scene, on interesting and informative regions.⁴⁰ The initial fixation on a target is sufficient to allow categorisation and identification judgements⁴¹ to allow judgements of trustworthiness²⁷ and provides sufficient information to correctly identify the general gist of the scene in front of the observer.⁴² Given this emphasis on data extracted in the first fixation, further analyses of fixation durations (and fixation distance, see below) were therefore based on the first fixation made towards the pedestrian. The mean duration of first fixations to pedestrians was calculated for each participant and these values used in subsequent analysis. The distribution of these mean values for each participant was also approximately normal and parametric statistical tests have been used.

The first fixation tended to be of longer duration when the fixated pedestrian was on their own (mean = 515 ms) compared with in a group (mean = 417 ms, p < 0.001). The gender of the fixated pedestrian may have had an influence on duration of the first fixation, with a longer time spent fixating females (mean = 502 ms) compared with males (mean = 468 ms), although this difference was not significant (p = 0.08). No other significant effects were found (Table 3).

Table 3

Mean duration and standard deviation of first fixations on pedestrians by different variables

Variable		Mean duration of first fixation (ms)	Standard deviation (ms)	Significance
Light condition	Day	472	143	n.s.^a
Light condition	After-dark	479	132
Gender (observer)	Male	499	135	n.s.^a
Gender (observer)	Female	453	135
Gender (fixated pedestrian)	Male	468	132	Near p = 0.08^b
Gender (fixated pedestrian)	Female	502	186
Location on fixated pedestrian	Head	481	182	n.s.^b
Location on fixated pedestrian	Body	481	209
Group	Solitary	515	179	p < 0.001^b
Group	Group	417	144
Direction of travel	Towards	511	187	n.s.^c
	Stationary	472	506
	Away	451	129

Independent t-test

Dependent t-test

Repeated-measures ANOVA

3.3. How far away do we fixate other pedestrians?

Analysis of the distance at which other pedestrians were fixated uses the weighted distance estimate which accounts for the systematic underestimation of estimates by the primary coder (see Section 2 above). Figure 3 shows the distribution of mean distances at first fixation towards other pedestrians. The mean distance to pedestrians when first fixated was 14.0 m (s.d. = 2.6 m).

Figure 3

Distribution of mean distance of first fixations towards other pedestrians (N = 54). Note: bin labels on the abscissa identify the upper limit of a bin range, e.g. ‘12’ is the range >9 but ≤12 m

This distance tended to be greater during the daytime (mean = 15.0 m) compared with after-dark (mean = 13.3 m, p = 0.02). This was found also in the previous analysis (daytime =13.0 m, after dark = 8.9 m). The distance a pedestrian was first fixated was significantly shorter if they were on their own (mean =13.8 m) than if they were part of a group (mean = 14.4 m, p = 0.02). Female pedestrians tended to be fixated at a greater distance (13.7 m) than were males (mean = 13.2 m, p = 0.03). There was a suggestion that pedestrians walking towards the observer were fixated at a slightly shorter distance (mean =13.5 m) compared with pedestrians who were stationary or walking away from the observer (respective means = 14.3 m for both), although this difference did not reach significance (repeated-measures ANOVA, p = 0.07). The gender of the observer and whether the first fixation was to the head or body did not have a significant effect on the distance of fixation (Table 4).

Table 4

Mean distance and standard deviation of first fixations on pedestrians by different variables

Variable		Mean distance at first fixation (m)	Standard deviation (m)	Significance
Light condition	Day	15.0	2.11	p = 0.02^a
Light condition	After-dark	13.3	2.70
Gender (observer)	Male	14.4	2.99	n.s.^a
Gender (observer)	Female	13.6	2.08
Gender (fixated pedestrian)	Male	13.2	2.82	p = 0.03^b
Gender (fixated pedestrian)	Female	13.7	2.81
Location on fixated pedestrian	Head	13.4	2.94	n.s.^b
Location on fixated pedestrian	Body	13.3	2.98
Group	Solitary	13.8	2.49	p = 0.02^b
Group	Group	14.4	3.24
Direction of travel	Towards	13.5	3.03	Near, p = 0.07^c
	Stationary	14.3	3.56
	Away	14.3	2.90

Independent t-test

Dependent t-test

Repeated-measures ANOVA

3.4. Do we fixate on different parts (head vs. body) on different fixations?

The probability of a participant looking at the body of another pedestrian rather than the head on the first fixation is 0.51 (a difference also shown in Table 2) and increases slightly to 0.53 for subsequent fixations. This change is not significant (dependent t-test, p = 0.36) and does not suggest a tendency to fixate the head or body at different fixation points.

While these figures suggest similar tendencies to fixate the head or body, an area-weighted analysis would show a greater tendency to fixate the head than the body⁴³ as has been found by others.⁴⁴ This may be because the face contributes more information than does the body. Hahn et al.³⁰ examined recognition judgements in videos of people walking towards the camera and found that the body contributes to recognition judgements at larger distances but that the face dominates at shorter distances. Burton et al.⁴⁵ found that recognition was significantly affected by loss of the face in observed images but not by loss of the body or of gait.

4. Comparison with previous analysis

The original analysis²² estimated fixation distance from 1683 discrete fixations on 1128 pedestrians and fixation duration from 177 fixations. The current analysis estimated distance and duration from 5955 fixations towards 2496 pedestrians, the higher numbers being due to an increased number of route sections, and is expected therefore to lead to more precise and accurate estimates of the nature of fixations toward other pedestrians. Table 5 compares estimates of fixation characteristics from the two studies.

Table 5

Comparison of pedestrian fixation characteristics in previous²² and current analyses

	Fotios et al.²²	Current analysis
			Significant affects
Number of fixations on a specific pedestrian	Mean = 1.75 (s.d. = 0.9)	Mean = 2.4 (s.d. = 0.6)	Greater number of fixations on pedestrians walking away than towards
Fixation duration	Median = 480 ms (IQR = 400–640 ms)	Mean = 475 ms (s.d. = 124 ms).	Longer duration of fixation on people when solo than when in groups
Mean fixation distance	Median = 10.3 m (IQR 8.3–12.3 m)	Mean = 14.0 m (s.d. = 2.6 m).	Greater viewing distance in day (than after dark), for people in groups Rather than solo) and for females (rather than males)

s.d.: standard deviation; IQR: Inter quartile range

The critical characteristics are the distance and duration of fixations as these parameters affect the ability to extract visual information and, therefore, the conclusions drawn about changes in lighting. The current analysis indicated fixations took place at a longer distance (mean = 14.0 m) than the previous analysis (median = 10.3 m). One reason for this is that different methods were used for estimating distance. In the first analysis²² distance was estimated by assuming each person fixated was of an average height (1.7 m). The current approach was to estimate distance by using as a cue the size of objects in the field of view and then weighting these estimates with a correction factor to counter the distance underestimate found in the second validation process.

This revised estimate provides further support for the proposal that 15 m is the distance at which it would be appropriate to investigate the effects of lighting on interpersonal judgements. It is a practical distance, being a shorter distance than that (23 m) at which the rate of correct facial identity recognition reduces to 25%,⁴⁶ and falls within the ‘action space’ zone (2–30 m) of Cutting and Vishton.²³ It is longer than the distances (4 m and 10 m) adopted by Caminada and van Bommel⁶ but agrees better with opinion from design guidance texts which propose there should be an ability to have a good look at other people at distances from 12 m to 25 m,^47–49 and agrees with Townshend’s finding of the preferred comfort distance after dark (15 m).⁵⁰ For 72% of the eye-tracking sessions analysed here, first fixations were made at a distance of less than 15 m.

The current analysis indicated fixations took place for a nearly identical duration (mean = 475 ms) to that found in the previous analysis (median 480 ms) and that (500 ms) which can be estimated from other work.⁵¹ For convenience we suggest this estimate of typical fixation duration is rounded to 500 ms. The duration of fixation was affected by group size, tending to be of longer duration when fixating a pedestrian on their own compared with in a group.

5. Conclusion

The aim of this paper is to identify the typical characteristics of fixations on other pedestrians and thus to inform the design and interpretation of experiments investigating how changes in lighting affect interpersonal judgements. Fixation characteristics were established through a review of the eye-tracking records of pedestrians walking outdoors in a natural setting in daytime and after dark, extending a previous analysis²² by analysing a much larger set of fixations. For future investigations of interpersonal judgements it is concluded that targets should simulate observation at a distance of 15 m and observation periods of 500 ms. The data do not indicate a need to discriminate between fixation on the head or body but that there may be a need to consider characteristics such as whether the fixated person is male or female and whether solo or in a group,

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Engineering and Physical Sciences Research Council (EPSRC) grant number EP/M02900X/1.

References

Fotios

Uttley

Yang

. Using eye-tracking to identify pedestrians’ critical visual tasks. Part 2. Fixation on pedestrians. Lighting Research and Technology 2015; 47: 149–160.

Foulsham

Walker

Kingstone

. The where, what and when of gaze allocation in the lab and the natural environment. Vision Research 2011; 51: 1920–1931.

British Standards Institution BS 5489-1:2013. Code of Practice for the Design of Road Lighting Part 1: Lighting of Roads and Public Amenity Areas, London: BSI, 2012.

Willis

Palermo

Burke

. Judging approachability on the face of it: The influence of face and body expressions on the perception of approachability. Emotion 2011; 11: 514–523.

Simons RH, Hargroves RA, Pollard NE, Simpson MD. Lighting criteria for residential roads and areas: Proceedings of the 21st Session of the CIE, Venice, 17–25 June 1987, pp. 274–277.

Caminada

van Bommel

WJM

. New lighting criteria for residential areas. Journal of the Illuminating Engineering Society 1984; 13: 350–358.

Boyce

Rea

. Security lighting: effects of illuminance and light source on the capabilities of guards and intruders. Lighting Research and Technology 1990; 22: 57–79.

Knight

. Field surveys investigating the effect of lamp spectrum on the perception of safety and comfort at night. Lighting Research and Technology 2010; 42: 313–330.

Knight C, Van Kemenade J. Wahrnehmung und Wertschätzung der Endkunden in Bezug auf die öffentliche Beleuchtung (Effect of outdoor lighting on perception and appreciation of endusers): Proceedings of Licht 2006, Berne, Switzerland, September 10-13: 2006.

10.

Johansson M, Rahm J. Perceived lighting qualities and pedestrian performance: Proceedings of 28th Session of the CIE, Manchester, UK, 28 June to 4 July: 2015: pp. 324–333.

11.

Rea

Bullough

Akashi

. Several views of metal halide and high pressure sodium lighting for outdoor applications. Lighting Research and Technology 2009; 41: 297–314.

12.

Yao

Sun

Lin

. Research on facial recognition and color identification under CMH and HPS lamps for road lighting. Leukos 2009; 6: 169–178.

13.

Alferdinck JWAM, Hogervorst MA, Van Eijk AMJ, Kusmierczyk JT. Mesopic Vision and Public Lighting – A Literature Review and a Face Recognition Experiment. Publication TNO-DV C435. Soesterberg, The Netherlands: TNO, 2010.

14.

Dong

Fotios

Lin

. The influence of observation duration and procedure on luminance required for recognition of pedestrian’ faces. Lighting Research and Technology 2015; 47: 693–704.

15.

Fotios

Yang

Cheal

. Effects of outdoor lighting on judgements of emotion and gaze direction. Lighting Research and Technology 2015; 47: 301–315.

16.

Fotios S, Castleton H, Cheal C, Yang B. Investigating the chromatic contribution to recognition of facial expression. Lighting Research and Technology. First published November 24 2015, doi:1477153515616166.

17.

Lin

Fotios

. Investigating methods for measuring facial recognition under different road lighting conditions. Lighting Research and Technology 2015; 47: 221–235.

18.

Yang

Fotios

. Lighting and recognition of emotion conveyed by facial expressions. Lighting Research and Technology 2015; 47: 964–975.

19.

Iwata M, Ayama M, Mori T, Iwasaki H, Kohko S, Inoue Y, Itsuki H, Kyoto N. Appearance of human face and atmosphere of environment under LED street lights using different correlated colour temperature: Proceedings of 28th session of the CIE, Manchester, UK, 28 June – 4 July 2015, 1638–1646.

20.

Rombauts

Vandewyngaerde

Maggetto

. Minimum semicylindrical illuminance and modelling in residential area lighting. Lighting Research and Technology 1989; 21: 49–55.

21.

Hall

. The Hidden Dimension, New York: Anchor Books, 1969, pp. 123–124.

22.

Fotios

Yang

Uttley

. Observing other pedestrians: Investigating the typical distance and duration of fixation. Lighting Research and Technology 2015; 47: 548–564.

23.

Cutting

Vishton

Perceiving layout and knowing distances: The integration, relative potency, and contextual use of different information about depth. In: Epstein

Rogers

(eds). Perception of Space and Motion, London: Academic Press, 1995, pp. 69–117.

24.

Adams

Zuckerman

. The effect of lighting conditions on personal space requirements. Journal of General Psychology 1991; 118: 335–340.

25.

Fujiyama

Childs

Boampong

Tyler

. Investigation of Lighting Levels for Pedestrians – Some Questions About Lighting Levels of Current Lighting Standards, London: Centre for Transport Studies, 2005.

26.

Cook

. Gaze and mutual gaze in social encounters. American Scientist 1977; 65: 328–333.

27.

Todorov

Pakrashi

Oosterhof

. Evaluating faces on trustworthiness after minimal time exposure. Social Cognition 2009; 27: 813–833.

28.

Willis

Todorov

. First impressions: Making up your mind after a 100-ms exposure to a face. Psychological Science 2006; 17: 592–598.

29.

Fotios

Cheal

Boyce

. Light source spectrum, brightness perception and visual performance in pedestrian environments: A review. Lighting Research and Technology 2005; 37: 271–294.

30.

Hahn

O’Toole

Phillips

. Dissecting the time course of person recognition in natural viewing environments. British Journal of Psychology 2016; 107: 117–134.

31.

Fotios

Uttley

Cheal

Hara

. Using eye-tracking to identify pedestrians’ critical visual tasks. Part 1. Dual task approach. Lighting Research and Technology 2015; 47: 133–148.

32.

Holmqvist

Nyström

Andersson

Dewhurst

Jarodzka

Van de Weijer

. Eye Tracking: A Comprehensive Guide to Methods and Measures, Oxford: Oxford University Press, 2011.

33.

Hallgren

. Computing inter-rater reliability for observational data: an overview and tutorial. Tutorials in Quantitative Methods for Psychology 2012; 8: 23–34.

34.

Landis

Koch

. The measurement of observer agreement for categorical data. Biometrics 1977; 33: 159–174.

35.

Cicchetti

. Guidelines, criteria, and rules of thumb for evaluating normed and standardized assessment instruments in psychology. Psychological Assessment 1994; 6: 284–290.

36.

Norman

Adkins

Pedersen

. The visual perception of distance ratios in physical space. Vision Research 2016; 123: 1–7.

37.

Waller

Richardson

. Correcting distance estimates by interacting with immersive virtual environments: Effects of task and available sensory information. Journal of Experimental Psychology: Applied 2008; 14: 61–72.

38.

Laidlaw

KEW

Foulsham

Kuhn

Kingstone

. Potential social interactions are important to social attention. Proceedings of the National Academy of Sciences 2011; 108: 5548–5553.

39.

Pilz

Vuong

Bülthoff

Thornton

. Walk this way: Approaching bodies can influence the processing of faces. Cognition 2011; 118: 17–31.

40.

Henderson

. Human gaze control during real-world scene perception. Trends in Cognitive Sciences 2003; 7: 498–504.

41.

Grill-Spector

Kanwisher

. Visual recognition: As soon as you know it is there, you know what it is. Psychological Science 2005; 16: 152–160.

42.

Underwood

. Cognitive Processes in Eye Guidance, Oxford: Oxford University Press, 2005.

43.

Fletcher-Watson

Findlay

Leekam

Benson

. Rapid detection of person information in a naturalistic scene. Perception 2008; 37: 571–583.

44.

Birmingham

Bischof

Kingstone

. Gaze selection in complex social scenes. Visual Cognition 2008; 16: 341–355.

45.

Burton

Wilson

Cowan

Bruce

. Face recognition in poor-quality video: Evidence from security surveillance. Psychological Science 1999; 10: 243–248.

46.

Loftus

Harley

. Why is it easier to identify someone close than far away? Psychonomic Bulletin and Review 2005; 12: 43–65.

47.

Colquhoun

. Design Out Crime: Creating Safe and Sustainable Communities, Oxford: Architectural Press, 2004.

48.

Dravitzki VK, Cleland BS, Walton D, Laing JN. Measuring commuting pedestrians’ concerns for personal safety and the influence of lighting on these concerns: Proceedings of the 26th Australasian Transport Research Forum, Wellington, New Zealand, 1–3 October 2003.

49.

Luymes

Tamminga

. Integrating public safety and use into planning urban greenways. Landscape and Urban Planning 1995; 33: 391–400.

50.

Townshend

The role of public lighting. In: Oc

Tiesdell

(eds). Safer City Centres: Reviving the Public Realm, London: Paul Chapman Publishing Ltd., 1997, pp. 119–129.

51.

Jovancevic-Misic

Hayhoe

. Adaptive gaze control in natural environments. Journal of Neuroscience 2009; 29: 6234–6238.