Effects of Time Exposure and Focused Attention on Person Bias in Preferential Looking

4124 words (16 pages) Essay in Psychology

23/09/19 Psychology Reference this

Disclaimer: This work has been submitted by a student. This is not an example of the work produced by our Essay Writing Service. You can view samples of our professional work here.

Any opinions, findings, conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of UK Essays.

The Effects of Time Exposure and Focused Attention on Person Bias in Preferential Looking

Abstract

Research on attention capture has studied whether representations of humans enjoy a preference in visual recognition for many decades.  Prior theories found that a preference does exist albeit with some counter evidence from alternative studies.  This study builds on prior work to determine if a viewing task can nullify the bias towards people in the fixation of attention.   The study also seeks to establish whether there is an inherent preference for human representations evidenced in its appearance in short presentation times.  The experiment also examines the interaction of viewing tasks and varied timing conditions on the preference for human representations.  Finally, this study investigates whether the CrowdGaze measure of visual fixation using letter and number pairs can replicate results measured through saccade-determining software. Preferential looking experiments to determine person present preference was tested on 98 Birkbeck College psychology students by replicating and building on experiments of prior studies.  Means analysis indicated that viewing task conditions showed more person fixations under shorter time conditions.  Results of a Two Way Mixed ANOVA showed that the probability of fixating on a person under viewing task is higher in the 200ms time condition than in the 1200ms condition.  However, the inherent preference for human representations based on time conditions studies is called into question by the non-significant nature of the relationship found here.  CrowdGaze appeared to replicate the results of earlier saccade-determined studies.  The mixed nature of the results in this study both supports prior theories of a preference for human representations in attention capture and runs counter to the hypothesis that these preferences are inherent.  Given these results further study on time conditions with more conditions and with a broader range should be undertaken.

Introduction

Whether representations of humans and in particular human faces enjoy a preferential capture of attention in visual recognition has been the object of study in psychological research for decades.  Early experiments in visual recognition identified that human figures versus other objects were disproportionately fixated upon evidenced by eye movements (Buswell, 1935).

In the 84 years since Buswell’s experiments, researchers have sought to understand the preferential nature of attentional capture through a body of preferential looking experiments and many have sought to replicate his findings.  One vein of research found a preference for animal representations, both human and otherwise.  Rousselet, Mace and Fabre-Thorpe in 2003 found evidence of an ultra-rapid attentional preference for animals and human faces though they did not detect a clear preference between the two (Rousselet et al, 2003).  Then in 2006, Kirchner and Thorpe found that subjects in their experiment reliably made eye movements to the side of split images containing animals, indicating attentional preference for animal figures (Kirchner & Thorpe, 2006).  They further concluded that low level differences did not account for the fast responses towards the animal figures, supporting the attentional preference for the animal figures (Kirchner & Thorpe, 2006).

Further inquiry looked specifically at the importance of human faces in preferential attention capture (Herschler & Hochstein, 2005, 2006) and found that face stimuli ‘popped out’ from other stimuli in visual search experiments.  Herschler and Hochstein also found that human facial stimuli captured attentional preference more than animal faces and more than separate facial elements alone (Herschler & Hochstein, 2005).  They confirmed their findings in 2006 (Herschler & Hochstein, 2006), refuting challenges to their results by Rufin VanRullen (VanRullen, 2006) who claimed that facial preference was not holistic and is due to a low-level confound.

Other studies have also challenged the preference for human faces in attentional capture.  In 1997, Brown, Huey and Findlay found no direct, rapid ‘pop out’ effect for human faces though they did show a processing advantage for upright faces over inverted faces in peripheral vision (Brown et al, 1997). 

Overall the trend in research has supported the preferential attention given to human faces in visual search experiments which was reinforced in the seminal study by Fletcher-Watson, Findlay, Leekam and Benson in 2008.  In that study the experimenters undertook experiments to replicate the findings of earlier studies that reinforced human attentional preference (Buswell, 1935; Herschler & Hochstein, 2005, 2006; Rousselet et al, 2003) and refute those studies challenging human preference (Brown et al, 1997; VanRullen, 2006).  Fletcher-Watson et al specifically wanted to determine if the preference for animal stimuli found in Kirchner & Thorpe (2006) could be replicated for humans and whether task-directed viewing instructions would have an influence on preferential attention configurations.  The Fletcher-Watson et al study (Fletcher-Watson et al, 2008) employed a split screen presentation with one half containing an image with a person present and the other half without a person.  They used a viewing task to determine the gender of the person on one half of the image to examine the effects of endogenous attention control and a free-viewing condition to examine exogenous attention control and the associated importance of human figures in prompting attentional preference (Fletcher-Watson et al, 2008).  Attentional preference was measured in this study using a saccade-detection programme.

The results of the Fletcher-Watson et al study showed “a strong bias towards attending to a scene containing a person, particularly towards that person” (Fletcher-Watson et al, 2008, p. 580).  They established that human faces attracted disproportionately more visual fixations, which proved exaggerated in the gender determination task, supporting the theory that human bodies and faces are subject to privileged perceptual processing.  They demonstrated that in both a task-directed condition and in a free-viewing condition there was a pronounced preference for the scene where a person was present; over 50% more in the gender determination task and a minimum of 7% more in the free-viewing condition (Fletcher-Watson et al, 2008). 

Importantly, Fletcher-Watson’s findings were consistent with Buswell, Herschler & Hochstein and Rousselet et al showing the preference for human figures in attention capture using visual search experimental methods.  The results replicated the animal preference findings of Kirchner & Thorpe (2006) for humans and the Fletcher-Watson et al gender determination task results also corroborate the findings of Smilek, Dixon and Merikle (2006) which held that category membership of objects has an influence on the efficiency of visual search.

Although Fletcher-Watson et al substantiated prior theories of preferences for human representation they did not specifically investigate the effect of presentation exposure times on attentional attraction nor did they examine the effect of a task direction to purposefully ignore the presence of a person in the images presented.  As well, Fletcher-Watson et al (2008) stopped short of considering the effect of the person in the image as a central figure warranting preferential attention versus any inherent predisposition towards human faces or bodies in general, as examined by Rensink, O’Regan and Clarke in 1997.  Further research on these considerations would add significantly to the literature.

The current experiment replicates in part the Fletcher-Watson et al (2008) experiments.  It seeks to build on their work to examine the main effects of two contrary tasks, one where subjects are told to observe the human representation in image trials and one where they are instructed not to observe it; specifically to determine if the second viewing task can nullify the bias towards people in the fixation of attention.   The study also seeks to establish the principal effect of the presentation time of images; particularly whether there will be an automatic preference for human representations in short presentation times.  The experiment further tests the interaction of viewing tasks and varied timing conditions on the preference for human representations.  Finally, this study investigates whether the CrowdGaze measure of visual fixation using letter and number pairs in a grid (Rudoy, Goldman, Shechtman & Zelnik-Manor (2012) can replicate the Fletcher-Watson et al (2008) findings on attentional preference which were determined using saccade-detection software.  In examining these hypotheses, the current study addresses the open questions of suitability of the CrowdGaze method and whether a viewing task can nullify the bias towards people in fixation of attention.

Method

Design

This experiment consisted of two distinct elements which mixed both a within subjects and a between subjects design.  The dependent variable in this experiment was the probability of fixating on the side of the image containing a person.  There were two independent variables each with two levels.  The first independent variable was the viewing task, one level represented by a gender determination element of the task and the second level by the instruction to avoid looking at people in the presented image trials.  The second independent variable was the manipulation of presentation times.  Level one was half of the 42 image trials presented for 200ms and level two had half of the images presented for 1200ms.

In this experiment subjects were asked to complete two separate tasks.  Task One was a viewing task replicating the Fletcher-Watson gender determination task (Fletcher-Watson et al, 2008).  Task Two was a viewing task designed to determine if instructing subjects to avoid looking at the people in the images would eradicate attention bias toward people.  A between-subjects design was used for these tasks; half of subjects completing one task and half completing the second. 

All subjects were shown the same 42 experimental image trials.  To test the effect of exposure time on preferential looking, 21 of the images remained on the screen for 200ms and 21 were exposed for 1200ms.  As all subjects experienced the same exposure time selected by the experimenter for each of 42 image trials, the time exposure test utilised a within subjects design. 

Determination of preferential looking was measured by the CrowdGaze method using number and letter combinations on slides presented immediately after flashing of each of the 42 image trials and asking subjects to record which combination they saw most clearly.  Using these data the probability of fixating on the side of the image including a person was calculated.  It was first determined if the letter and number pairs recorded by each subject for each image was on the person-present side or not.  Then the average across 21 image trials for each presentation time condition was then determined to yield a probability between 0 and 1.

 

Participants

The subjects were Birkbeck College psychology students participating in the experiment as a routine part of their study.  The sample was 98 adult students; 72 females and 26 males between 19 and 55 years of age (N = 98, mean age = 30.24, St. dev. = 8.83).  Subjects were verbally informed of their right not to participate in the experiment or to stop at any time.  Subjects were informed that all data collected would be anonymous.

 

Materials

Subjects had a clear view of experimental slides in PowerPoint projected onto a screen in front of the lecture hall.   Subjects were randomly assigned subject numbers and were well-spaced to avoid distraction.  Written instructions which included a data capture section were distributed to subjects.  Forty-six of the subjects received a set of instructions describing Viewing Task One “Gender Discrimination.”  Fifty-two subjects received instructions for Viewing Task Two “Avoid Looking at the People.”  Both data capture/instruction sheets queried subjects’ age, gender, native language, handedness, whether subjects were wearing glasses or contacts and where in the lecture hall they were seated (see Appendix for example instruction/data capture sheets).  The projected slides included a series of 42 images.  Each image was comprised of two side-by-side pictures containing a mixture of various rooms, either including a person looking at an item or not.  The slide set also included a black background slide containing white letter/number combinations in a grid formation (see Appendix) which were placed after the split image slides.

Procedure

All subjects received instruction/data capture sheets for the experiment.  For Task One “Gender Determination” subjects were instructed to view each of a series of images followed by a grid of letter and number pairs.  After each image they were to record the gender of the person in the image, if the image contained one, and then record which letter and number pair appeared most clear to them. 

For Task Two “Avoid Looking at the People” subjects were also instructed to view the images presented followed by the letter and number pairs.  They were instructed to avoid looking at people in the images, if one was presented, and similarly to record which pair was most clear to them.  Task Two was completed by half of the subjects contemporaneously with the other half of subjects completing Task One, all presented with the same images and grids.  All subjects participating in both tasks were provided with two demonstration trials and one practice trial to familiarise them with the procedure of the experiment.

Once the experiment was initiated, subjects were presented with a black slide containing one white cross in the centre to direct their gaze to the centre.  This was presented for 500ms.  Immediately after, subjects were presented by each of the 42 image trials, 21 presented for 200ms and 21 presented for 1200ms, randomly assigned.  Once each image disappeared, a black slide with the letter and number pairs were presented for 500ms after which subjects recorded on data sheets the data they were instructed to capture.

Results

To test the effects of the two time conditions on attention bias towards people, the effects of instructing subjects to attend or avoid representations of people in stimuli and interactions among the independent variables on the dependent variable of probability of fixating on people in the stimuli, a mixed 2×2 ANOVA was performed.  A Levene’s test was run to determine whether the data passed the homogeneity test in order to rely on the ANOVA.  Mauchly’s test was not used as the independent variables had only two levels.  The results of the Levene’s test were not significant (presentation time 200ms, F = .665, p = .417 and presentation time 1200ms, F = .547, p = .461) therefore assumptions are met that the within subjects groups we are comparing have a similar dispersion of scores.

Before turning to the ANOVA results it is useful to consider the shape of the data for each of the time conditions and task groups.  Descriptive statistics showed different characteristics among these conditions albeit within an overall small range.   As a whole, subjects in the two viewing tasks showed more fixation on the person-present image  under the 200ms time condition (M = .534, St. dev. = .300) than in the 1200ms condition (M= .453, St. dev. .333).  On average, subjects conducting the Gender Discrimination task performed similarly under the 200ms time condition (M = .767, St. dev. = .203) and the 1200ms condition (M = .719, St. dev. = .222).  However, subjects conducting the Avoid People task performed better in the 200ms condition (M = .328, St. dev. = .206) than in the 1200ms condition (M = .218, St. dev. = .218).  Table 1 below summarises the descriptive statistics for this experiment.

Table 1: Descriptive Statistics Results

Time condition

Viewing task

Mean

Std. Deviation

N

Presentation time 200ms

Gender discrimination

.767

.203

46

Avoid people

.328

.206

52

Total

.534

.300

98

Presentation time 1200ms

Gender discrimination

.719

.222

46

Avoid people

.218

.218

52

Total

.453

.333

98

The results of the Two-Way Mixed ANOVA showed that there was a significant effect of time condition on attention to the person-present sides of the image trials overall

(F (1,96) = 21.53, p = < .001, p2 = .183).  In addition, the test showed no significant interaction between time conditions and performance of the 2 task groups performing either the Gender Determination or the Avoid People task (F (1,96) = 3.28, p = .073, p2 = .033).  The ANOVA also indicated that there was a significant effect of the task performed with respect to the person-present bias of preferential looking during the experiment (F (1,96) = 141.50, p = < .001, p2 = .596).

A graphical representation of the ANOVA results demonstrates these relationships among all four independent variables and the probability of fixating on the side of the image including a person.  This is presented Figure 1, below.

Figure 1: The Effects of Viewing Tasks and Time Conditions on Probabilities of Person/Face Fixation

Figure 1 shows that the probability of fixating on a person or face in the Gender Determination task is much higher than in the Avoid People task and that in both tasks the fixation is more probable to occur in the 200ms time condition than in the 1200ms condition.  It also demonstrates that probabilities decrease more markedly in the Avoid People task between the two time conditions than in the Gender Determination task.

Discussion

The results demonstrate support for the theories of a human representation preference in attentional capture.  The descriptive statistics show more fixation on the human representation side of image trials in the 200ms condition than in the 1200ms condition suggesting that there is an inherent preference for human representation since extended timer exposure showed no processing advantage.  While the person-present preference in the task requiring focus on gender was only slightly higher in the 200ms time condition than in the 1200ms condition, it was much higher in the Avoid People task where subjects were instructed not to attend to the human representation; all of which supports the theories of Fletcher-Watson et al (2008) and prior studies regarding person preference in attention capture.

The results further show that the effect of time condition on attention to human representation described above was indeed significant, buttressing the hypothesis that human representation enjoys a preference in attention capture.  In addition, the effect of the task performed was found to be significant, suggesting that the instruction to attend or not to attend to human representation is an important factor in where attention is directed.  Unsurprisingly, the results also indicate a much higher probability to fixate on the person present when instructed to do so, even though there was a person present preference in both tasks.  These results reinforce the findings of Buswell (1935), Fletcher-Watson et al (2008), Herschler & Hochstein (2005, 2006) and Rousselet et al (2003) supporting human representation preference in attention. 

However, there is a counter indication present in these results since the interaction between time conditions and the person present preference in either task condition was not found to be significant, suggesting that there may not be an inherent preference for human representations in attention capture.  Given the weight of prior theories in support of the inherent preference, further research on the effect of time conditions would be warranted.

The current study sought to replicate in part the experiments of Fletcher-Watson et al (2008) which found human preference in attention capture measured by eye movements, in their case using saccade-determining software.  Those results were replicated here using the CrowdGaze method which appears to support the efficacy of that method of measurement.  This study also sought to determine if the Avoid People task could nullify the bias towards human representations in attention capture.  Results indicate that it did not since both tasks were found to be significant factors in the person present preference.  This experiment investigated whether there would be an automatic preference for human representations based on the presence of the preference in short time conditions.  Though results showed this to be the case prima facie in the descriptive statistics, that relationship was found not to be significant.  Consequently, the question of an inherent person present preference corroborated by time condition study remains open.

The mixed nature of the results here both supports prior theories of a preference for human representations in attention capture and runs counter to the hypothesis that these preferences are inherent due to the non-significant effect of time conditions.  Given that the inherency of the human representation preference remains an open question, further study on time conditions with more conditions and with a broader range should be undertaken.

Notwithstanding the qualified support for the experimental hypotheses, this experiment had a series of limitations that question reliance on the results.  The range and number of time conditions was limited calling into question the reliability of the results on the effects of time conditions.  Another limitation was that this experiment only tested two tasks and failed to include a free-viewing task as did Fletcher-Watson et al (2008).  Including additional tasks in subsequent research would add to the investigation of person present preferences in attention capture.  This experiment did not account for the possibility of the person as a central figure in the images and its effect on the person preference which could be addressed by further experimentation.  In addition, though this one experiment appears to support the efficacy of the GrowdGaze method of visual attention capture, expansion of this research is warranted to produce a reliable result.  This wider exploration could advance the examination of human representation reference in attention capture and the inherency of it.

Conclusion

The experimental results both support prior theories of a person present preference in attention capture and fail to support the hypothesis that these preferences are inherent due to the non-significant effect of time conditions.  Since the inherency of the human preference remains an open question, further experimentation on the true nature of  human presence preference in attention capture would be warranted and add to the body of research. 

References

  • Brown, V., Huey, D., & Findlay, J. M. (1997). Face detection in peripheral vision: Do faces pop out? Perception26(12), 1555-1570.
  • Buswell, G. T. (1935). How people look at pictures: a study of the psychology and perception in art.
  • Fletcher-Watson, S., Findlay, J. M., Leekam, S. R., & Benson, V. (2008). Rapid detection of person information in a naturalistic scene. Perception37(4), 571-583.
  • Hershler, O., & Hochstein, S. (2005). At first sight: A high-level pop out effect for faces. Vision research45(13), 1707-1724.
  • Hershler, O., & Hochstein, S. (2006). With a careful look: Still no low-level confound to face pop-out. Vision research46(18), 3028-3035.
  • Kirchner, H., & Thorpe, S. J. (2006). Ultra-rapid object detection with saccadic eye movements: Visual processing speed revisited. Vision research46(11), 1762-1776.
  • Rensink, R. A., O’Regan, J. K., & Clark, J. J. (1997). To see or not to see: The need for attention to perceive changes in scenes. Psychological science8(5), 368-373.
  • Rousselet, G. A., Macé, M. J. M., & Fabre-Thorpe, M. (2003). Is it an animal? Is it a human face? Fast processing in upright and inverted natural scenes. Journal of vision3(6), 5-5.
  • Rudoy, D., Goldman, D. B., Shechtman, E., & Zelnik-Manor, L. (2012). Crowdsourcing gaze data collection. arXiv preprint arXiv:1204.3367.
  • Smilek, D., Dixon, M. J., & Merikle, P. M. (2006). Revisiting the category effect: The influence of meaning and search strategy on the efficiency of visual search. Brain Research1080(1), 73-90.
  • VanRullen, R. (2006). On second glance: Still no high-level pop-out effect for faces. Vision research46(18), 3017-3027.

Cite This Work

To export a reference to this article please select a referencing stye below:

Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.

Related Services

View all

DMCA / Removal Request

If you are the original writer of this essay and no longer wish to have the essay published on the UK Essays website then please: