Printer Friendly

Effects of a large area glare source in cognitive efficiency and effectiveness in visual display terminal work.

1 INTRODUCTION

Since the early 1980s, the study of visual display terminal (VDT) work has focused on visual comfort and postural problems. The introduction of information and communication technologies (ICT) in offices has added a constant cognitive processing load [Wastlund 2007]. The hypothesis behind this investigation is that glaring lighting conditions might be related to nonvisual effects affecting ICT operators' cognitive efficiency and effectiveness. We designed an experiment to test the hypothesis that glaring scenarios could generate not only visual strain but cognitive strain as well. This research is in line with the "third age of lighting research" [Cuttle 2010] which is characterized by the study of nonvisual effects of light such as the regulation of the human circadian rhythms [Fonseca and others 2006] or the lighting effects in cognitive performance [Baron and others 1992, Knez 1995].

2 THEORETICAL FRAMEWORK

In Masuda's best-known book "The Information Society and Post-Industrial Society" [Masuda 1980], he built the concept of an information society (IS), which has become a reality. As with other technological revolutions of mankind, IS has penetrated into every domain of human activity introducing discontinuous patterns in the economy, society, and culture. ICTs and the Internet are engines behind that revolution [Kranzberg and Pursell 1981].

Those transformations have changed clerical work, and the concept of ICT worker arises. This paper redefines clerical work with VDTs, considering the cognitive demands that computer systems impose to the worker as he or her performs daily activities. Neither descriptive nor normative studies of lighting have yet addressed this subject. The theoretical framework for office lighting built in the last 30 years focused on the visual aspects of VDT work but not the cognitive questions derived from the introduction of ICT in this context. A brief summary follows.

The electronic office introduced in the 1980s brought new concerns in lighting design [Blehm and others 2005], mainly the prevention of disability and discomfort glare in VDT operators [Osterhaus 1993]. The "Vocabulaire international de reclairage" [CIE 1987] defined glare as the particular condition that could cause discomfort or could reduce visual performance, the visibility and the capability to define details and objects, caused by an unsuitable luminance distribution, or by high luminance contrasts within the visual field. Disability glare is the effect associated with reductions in visual performance, but not necessarily coupled with discomfort sensations. Discomfort glare refers to the sensation perceived, which is not necessarily tied to a reduction in visual performance. A strong body of knowledge has been developed around the assessment of visual comfort, associated to the absence of indoors discomfort glare [Bellia and others 2008] from artificial and natural light sources.

Natural light, controlled, has positive impact on human health and performance, as well as on thermal and lighting efficiency of constructed spaces [Boyce and others 2003]. It is also preferred as a light source [Galasiu and Veitch 2006] and provides a view to the outdoors [Tuaycharoen and Tregenza 2007]. If glare from windows is avoided, the energy consumption associated with the electric lighting may be reduced and lighting quality may be improved. These considerations are part of lighting quality [Veitch and Newsham 1996], which meets not only economic requirements, but also psychological [Van Bommel 2006] and biophysical needs [Hoffmanna and others 2008]. These advantages make daylighting systems a favorite of designers and engineers. Glare, however, often correlates with daylighting through high or nonuniform luminance distribution within the visual field or high luminance contrast between a window and its surroundings. Past studies have demonstrated that discomfort glare from windows depends upon the position and size of the glazing, as well as on the part of sky seen through it (see Bellia and others [2008] for a review). For computer tasks, the disadvantage could be both by veiling glare and by a high contrast between the luminance of the background and the VDT.

The main communication channel between an operator and ICTs is visual, through a VDT [Koch and Prinz 2002], involving basic cognitive processes such as attention, perception, memory, decision-making and stimuli interpretation. The ICTs provide a possibility for quick and direct interaction between users and give access to enormous quantities of information. Biocca [2000] discusses the prospect that the Internet will likely continue to grow at a rapid pace for at least another 10-20 years. Technostress is a term [Arnetz and Wikholm 1997] that describes the state of mental and physiological arousal observed in persons heavily dependent on computers in their work. One type of stressor involved in relation to ICTs is information overload, which raises mental demands [Wast-lund 2007] beyond the very limited human capabilities. Cowan [2010] states that there are severe limits in how much can be kept in one's mind at once, suggesting it is only about 3 to 5 items.

Information processing can be described and explored in terms of working memory (WM) [Duff and Logie 2001]. Working memory [Repovs and Baddeley 2006] plays a crucial role in every activity that requires temporal storage and management of information. WM is important in the use of the computer [Hashizume and others 2007]: it retains the incoming information and changes and renews the contents of information accordingly to the operation and the processing. WM is a memory tool for achieving a goal. The WM construct is an extension of an earlier concept: short term memory (STM), a unitary temporal storage system [Atkinson and Shiffrin 1968] capable of storing 7 [+ or -] 2 "chunks". In his classic paper, Miller [1956] defined a chunk as any piece of meaningful information such as letters, digits or words. The differences between WM and STM are: (i) WM has several sub systems and (ii) WM provides a functional basis for complex cognitive tasks such as learning, reasoning and comprehension. WM is defined as the memory function able to process and hold the information needed to perform a specific task, adaptable to the situational changes and the task progress itself [Hashizume and others 2007].

The first WM model [Baddeley and Hitch 1974] had three functional components: The central executive (CE) of limited attention capacity that handles information and controls two slave storing systems: the phonological loop (PL) for storing and holding verbal information and the visuospatial sketchpad (VSP) for storing and holding visual and spatial information. A fourth component was recently proposed: the episodic buffer of limited capacity capable of multidimensional coding and information integration of information to create integrated episodes [Baddeley 2000].

Several alternative working memory models have emerged (see Miyake and Shah [1999] for a review). A specific point of divergence pertains to how the very narrow span of the working memory can account for the large processing demands made by complex problem solving. Irrespective of the model, at the heart of the matter lies a short-term memory system activated for both short time retention and a central processing structure in the service of complex cognition.

The hypothesis behind this investigation is that glaring lighting conditions might be related to nonvisual effects affecting ICT operators' cognitive efficiency and effectiveness.

3 MATERIALS AND METHODS

In order to test our hypothesis we designed an experiment to determine whether or not glaring scenarios could generate not only visual strain, but cognitive strain as well. Glare caused by natural lighting is usually studied in dark chambers lit with devices emulating a window: the photometric independent variables measured are correlated with the subjective sensation of glare from the observers and turned into equations [Hopkinson 1972, Chauvel and others 1980, Iwata and others 1991]. Those kinds of experiments are the source behind many well-known glare indexes. The same strategy was used in the present study: the use of an emulated window in a laboratory to investigate glare not just as a visual stressor but also as a cognitive one.

3.1 INDEPENDENT VARIABLES

A full factorial experimental design was employed with three factors at two levels. The first factor was Luminance (L). By definition, high luminance or high contrasts of luminance in the visual field cause glare [CIE 1995]. The criterion to configure each treatment was to achieve L intensities that, theoretically, were glaring and not glaring. Conclusions from previous studies [Osterhaus 2002, NUTEK 1994] were a reference to set the lower level of L below 1.5 x [10.sup.3] cd/[m.sup.2] and a luminance ratio between task and window below 1:100, and the higher level above 1.5 x [10.sup.3] cd/[m.sup.2] with a luminance ratio above 1:100.

The second factor was the apparent source size in steradians (sr), defined as the subtended solid angle of the source at the eye of the subject. The lower level was the typical minimum size of a real window. Hopkinson [1972] asserted that that size is 0.01 sr. The human visual field has approximately 2[pi] sr and as the subtended angle reaches that size the source dominates visual adaptation, lowering the chances of glare [Hopkinson 1960]. The higher level for the source size factor considered that phenomenon and was set at 0.66 sr.

The third factor has its roots in cognitive psychology. In studies of individual differences of WM, dual task scenarios requiring active maintenance of attention in front of powerful environmental distractors have been employed. An appropriate distractor is the antisaccade task [Hallet 1978]. In this experiment, the light source was considered a "powerful environmental distractor". In the prosaccade condition, we asked the subject to change his visual field from the task to the light source for a second, allowing successive visual adaptation and accommodation changes and increasing the probability of glare. We believe the prosaccade condition has high ecological validity because it is likely that the subjects shift their vision between the VDT and the light source in actual scenarios. In the antisaccade condition, the subject had to ignore the distractor while performing the task. Although we reduced glare chances, more attention resources had to be assigned to avoid the stimulus, increasing the cognitive weight of the treatment. The prosaccade condition simply requires looking toward the light source and this response is thought to rely on exogenous attentional capture and should not require the recruitment of executive control. The antisaccade condition, however, requires the inhibition of a prepotent response and thus requires a degree of attention control not apparent in the relatively automatic prosaccades. Therefore, antisaccades but not prosaccades should require executive control. [Unsworth and others 2004]

3.2 DEPENDENT VARIABLES

This study had two dependent variables: cognitive effectiveness and cognitive efficiency. The relationship between them was described by De Waard [1996], who showed that performance measures (effectiveness) are less sensitive to workload variations than subjective assessments of the amount of cognitive resources allocated by the subject (efficiency).

The first dependent variable was cognitive effectiveness by means of performance in the reading span task (RST) [Daneman and Carpenter 1980]. The RST is a complex performance measure that requires both storage and information processing, in contrast to simple measurements that require only storage [Repovs and Baddeley 2006]. This requirement to both process and store information involves the central executive part of WM and provides a closer approximation to everyday complex cognitive tasks than simple storage [Daneman and Carpenter 1980]. The RST correlates with a wide range of high order cognitive tasks present in ICT work [Conway and others 2007]. Its self-consistency is 0.78 Cronbach's Alpha [Kane and others 2004].

[FIGURE 1 OMITTED]

In the original version of RST developed by Daneman and Carpenter [1980], subjects were required to read aloud, at their own pace, sentences presented on index cards and to remember the last word of each sentence for later recall. The sentences were presented in groups that ranged in size from two to six (Daneman and Carpenter referred to a group of sentences as an "item"). The subjects also had to indicate the veracity of each sentence by giving a true or false response within 1.5 sec of each sentence's presentation. The sentences were drawn from general knowledge quiz books and were selected to be of moderate difficulty. Although Daneman and Carpenter [1980] did not monitor the subjects' accuracy on the true-false component, the subjects believed it was an important part of the task. This prevented the subjects from adopting a strategy of focusing on the final words without devoting much attention to reading the sentences. There were 15 items, three each consisting of two, three, four, five, and six sentences that were 13-16 words in length. Increasingly larger groups of sentences were presented until the subject failed to recall all three items of a given size. At this point, the experiment was terminated. A subject's reading span was the level at which he or she could correctly recall two of the three items. For example, if a subject were to successfully recall at least two out of three two-word items, the experiment would continue for the subject to attempt three-word items. If the subject were then to successfully recall only one out of three of the three-word items, the experiment would terminate, and the subject's reading span would be two.

In this experiment, we used a modified Spanish version of RST. Instead of paper cards, we digitally presented the task to the subjects on a notebook screen through a MS PowerPoint presentation (Fig. 1 left). Each group of sentences appeared in white text on black background (mean luminance contrast = 2.21), in a Verdana 12 point font (subtended angle on the eye = 0.00006 sr), left aligned and centered on the screen. Each sentence was visible for six seconds and then disappeared. Thanks to a delay of one second between the slides, the subjects had the same time for reading in prosaccadic and antisaccadic treatments. The subject then had to press the spacebar to read the next sentence. At the end of each item a blue screen appeared, indicating to the volunteer that he had to say aloud the words he had in his memory. To move to the next item the subject had to press the spacebar. A total of 800 different sentences were generated with a length of eight to twelve words each and were randomly presented to each subject. The last word of the sentence, which should be remembered, varied from one to four syllables. The basic process to retrieve and re-articulate the contents held in the phonological store and refresh the memory trace is the articulatory rehearsal, a mechanism analogous to subvocal speech. To inhibit the rehearsal mechanism, the subject had to read aloud the sentences. To include processing besides storage, the subjects also had to indicate whether or not any word of the sentence had an accentuation error (for example, in Spanish: camion/camion). As in the Daneman and Carpenter [1980] experiment, we did not monitor the accuracy of the accentuation component and the subjects were not aware of that. Working memory span was the level at which the subject could correctly recall two of the three items. The experimenter had a sheet of paper with the to-be-remembered words and monitored the subject's answers to indicate whether or not the subject should continue to the next level.

The second dependent variable of the experiment was cognitive efficiency, defined as the amount of cognitive resources needed to perform a complex cognitive task. Self-report assessments have always been appealing to researchers because no one is able to provide a more accurate judgment about the experienced mental load than the person involved. Self-report scales have high face validity, are easy to apply and have low costs of application [O'Donnell and Eggemeier 1986]. The NASA task load index (TLX) [Hart and Staveland 1988] is a multidimensional scale that uses six dimensions to assess mental workload: mental demand, physical demand, temporal demand, performance, effort, and frustration. This procedure requires a weighting procedure to combine the six individual scale ratings into a global score. Byers and others [1989] proposed a raw task load index (RTLX) that does not require task paired comparison weights. The RTLX is a simple average of the six TLX scales. Byers and his colleagues found that TLX and RTLX had comparable means and standard deviations, and correlated above r = 0.95; they recommend the RTLX as a simple alternative to the TLX. Comparisons among self-report methods [Rubio and others 2004] found in TLX desirable psychometric properties: low intrusiveness and implementation requirements, and high sensitivity, diagnosticity, validity and user acceptance. Based on the high correlations between the traditional TXL and the raw TLX we decided to derive the overall workload ratings using the simpler and less time consuming RTLX method. We presented the RTLX form digitally on the computer screen (Fig. 1, right) with positive polarity as is usual for written documents in actual VDT work. RTLX was not employed as a task, but as an assessment method completed after the RST task was done. The RTLX form was employed to rate how difficult the subject felt it was to reach the level of RST achieved in each trial. No cofounding effects were expected for the different polarities between RST and RTLX.

We created a full factorial design and randomized the treatments with MINITAB 14 for Windows. For sample size determination, we used RTLX standard deviation from a previous study [Rodriguez and Pattini 2011]. The results showed a sample size of 400 runs, with 50 repetitions for each of the eight treatments.

The participation requirements were under 35 year of age, normal or corrected vision, and not being under any medical treatment. The experiment took place inside the experimental light laboratory at CCT CONICET-Mendoza Argentina. Figure 2 shows on the left the experimental set up. On the right of Fig. 2 is a view outside the laboratory and the VDT workstation.

[FIGURE 2 OMITTED]

A Compact Presario F700 notebook, TFT 14.2" widescreen (screen apparent size = 0.342 sr) presented the task and the digital forms. The workstation geometry was the same for every subject thanks to chair regulations and a chin rest to keep the angle and distance among the eye, the screen, and the large area source constant.

The large area source was an artificial window of 1.5 m wide and 1 m high. Two openings behind it improved ventilation and air exchange with the exterior. Inside the window, aluminum paper reflected the light generated by the multiple sources making the device more efficient. Actual windows have a nonuniform luminous distribution. The sky and the obstacles seen through them have different luminance; furthermore, the sky is not uniform. Thus, windows should be considered as large glare sources with nonuniform luminance [Kim 2008]. In order to create a nonuniform luminance glare source we randomly arranged fifty-four incandescent lamps in a 6x9 matrix. The power of the sources varied between 40 W and 150 W. A plastic screen avoided direct vision of the lamps and a cardboard shutter allowed changing the size of the source (from an apparent size of= 0.667 sr to an apparent size of 0.017 sr). Both sources were centered on the subject's horizontal line of vision (Fig. 2) so the angular shift between large and small source conditions was the same.

3.3 TREATMENTS CHARACTERIZATION

In actual workspaces the field of view (FOV) is centered in the task and, depending on the equipment layout, the glare source is somewhere around the task within the worker's FOV. Eventually the worker may look directly at the window, but most of the time he or she is working on the computer screen (and is adapted to its luminance). We performed a glare analysis to visually characterize each treatment with a task-centered criterion. We obtained values of daylight glare index (DGI) from luminance maps by means of high dynamic range images (HDRI) [Inanici and Galvin 2004]. There were two factors related to the source itself: its luminance and its size, whereas the third factor depended on if the subject shifted his or her visual field between the task and the source (prosaccade task) or not (antisaccade task).

[FIGURE 3 OMITTED]

A series of low dynamic range images (LDRI) were taken with a Nikon Coolpix 5400 camera with a Nikon FC-E9 Fish Eye lens. Each image was taken from the position of the subjects' eyes and pointing to the center of the VDT. The LDRIs were processed with Photosphere for MAC OS. As the information stored in HDRI corresponds with photometric values of luminance [Ward Larson and Shakespeare 1998], this technique replaces punctual measures with a luminance meter. We, however, used a Minolta LS100 luminance meter to obtain control luminances in the higher and lower extremes of the dynamic range to calibrate the scenes. We processed the HDR images with Evalglare software and calculated DGI for each treatment. The borderline between comfort and discomfort (BCD) [Tokura and others 1996] corresponds to 20 in the scale.

We assessed the visual strain that each treatment caused on the subject by means of glare sensation. Discomfort glare is essentially a subjective phenomena and demands research methods involving subjective assessment, although complemented with objective data [Osterhaus 2005]. The assessment method chosen was glare sensation vote (GSV) [Hopkinson 1972], which estimates the glare sensation as a function of the time the subject could stand the sensation of discomfort (Fig. 3). As the other assessment method used in this experiment, a digital form (positive polarity) presented the scale and included for each point a definition. For instance, "just intolerable " was defined as the situation where the subject could not stand the discomfort caused by the source and, if given the chance, would immediately take action in order to reach a state of comfort. This scale has been widely used since its introduction [Chauvel and others, 1980; Osterhaus and Bailey 1992; Kim and others, 2009]. The borderline between comfort and discomfort (BCD) [Kim and others 2009] corresponds to 1.5 in the scale.

We recorded every trial of the experiment using the notebook's webcam (Fig. 4). This allowed us to characterize where the subjects were looking and to register the duration of each trial. Figure 4 shows a prosaccadic condition on the left and an antisaccadic condition on the right.

We also monitored the room temperature with a LMT-8000 multiple environmental measurement device at the beginning, the middle, and the end of each experimental session. The mean temperature should be in the 27 [+ or -] 3 [degrees]C range to avoid cofounding effects [Hygge and Knez 2001].

[FIGURE 4 OMITTED]

4 RESULTS AND DISCUSSION

Data collection lasted 90 days. All the measurement sessions were in the morning. Each volunteer had a training run and two experimental runs. Figure 5 shows the sequence of the experiment. The training session was recorded with the notebook's web cam which allowed us to register the initial (ti.i) and final time ([ti.sub.f])of each session. The subjects performed four repetitions of RST and RTLX in a neutral lighting scenario where the window was switched off. Two of the repetitions were antisaccadic and the other two were prosaccadic. The aims of the training session were to achieve some proficiency on the RST and to avoid learning effects during the experiment. At the end of the training session, glare sensation was assessed by means of GSV.

In the training sessions, 31 volunteers performed the RST (mean age 30.16 years old; SD 3.95). The mean duration of the training sessions was 30 minutes, with a SD of 10.33. The sample had 33.5 percent men and 64.5 percent women. Whereas women generally outperform men in episodicmemory tasks, men and women do not differ significantly in any type of WM, men and women having almost identical WM architecture [Roberta and Savoie 2006]. No gender differences in glare sensation vote were found in previous studies [Iwata and others 1992] and no confounding effects were expected from this sample's gender distribution. 66.7 percent of the subjects had normal vision and the remaining 33.3 percent wore glasses during the experiment. While studying the influence of window view in glare sensation, Tuaycharoen and Tregenza [2007] found no glare assessment significant differences as a function of eyeglass usage.

Two days later, in the experimental run, the volunteers passed through the eight treatments (Ti to T8) of the factorial design which were randomized to avoid order effects. In each treatment, the subject performed the RST until his WM limit was reached. Then the volunteer completed the RTLX and GSV digital forms for that treatment while the experimenter prepared the following scenario.

[FIGURE 5 OMITTED] Once glare sensation and cognitive efficiency was assessed for that treatment, the next treatment began and the RST - GSV - RTLX sequence was repeated under a different lighting environment. Before the fifth scenario there was a ten-minute break to minimize fatigue effects on the last four scenarios. The session was video recorded, initial (tii) and final (tif) time was registered. Room temperature was controlled (to, t1, and t2) during the session.

[FIGURE 6 OMITTED]

Two of the trained volunteers chose not to be part of the experimental run, whereas four of them were dropped because the mean temperature in their runs was above 27 [+ or -] 3[degrees] C. The mean duration of the experimental sessions was 1:11 hour (SD = 21 min) with the shortest session of 44 minutes and the longest session of 2:04 hours There were a total of 400 trials, in 50 sessions. In each session the volunteer repeated the task 8 times.

Figure 6 shows false color luminance maps from Photosphere. The scale reaches 2* [10.sup.3] cd/[m.sup.2]. The images on the left show low luminance treatments and images on the right show high luminance treatments. The high level for the size factor is above and the low level is grouped below. Table 1 summarizes the values of luminance contrast measured from the HDRI and DGI from Evalglare and the mean subjective assessment (GSV) for each treatment. Sensations above BCD are highlighted in green. The high luminance treatments presented higher values of GSV, except for the prosaccadic/low/small combination. For each combination of source size and luminance, the subject had to shift his vision with the source. As a result, being under the same photometric setting, higher glare sensations were found in prosaccadic conditions in relation to antisaccadic conditions whereas calculated DGI remained constant. Discomfort glare depends not only on physical and photometric factors but also on temporal factors: how long and how often the glare source is in the FOV. Since equations do not yet include this temporal dimension, we calculated a priori DGI from the most representative (in terms of time) FOV (centered on the VDT) and assessed a posteriori glare sensations for the prosaccadic and antisaccadic conditions. Glare prediction should therefore include the likelihood of the light source actually occupying the visual field of the worker.
TABLE 1. Treatments Glare Characterization: GSV and DGI.

Luminance  Size             Task       GSV     [L.sub.s]
                                             (cd/[m.sup.2])

           Large  24.3  Prosaccade   3.12

                        Antisaccade  2.88          1.82 X
                                               [10.sup.3]

High       Small  23.9  Prosaccade   2.00

                        Antisaccade  1.60          1.91 X
                                               [10.sup.3]

           Large  17.6  Prosaccade   1.64

                        Antisaccade  1.26          2.83 X
                                               [10.sup.2]

Low        Small  15.6  Prosaccade   1.12

                        Antisaccade  1.48          3.37 X
                                               [10.sup.2]

Luminance    [L.sub.t]       Contrast
           (cd/[m.sup.2])

                   5.20 X        3.50 X
               [10.sup.0]  [10.sup.2]:1

High

                   6.40 X        2.99 X
               [10.sup.0]  [10.sup.2]:1

                   5.40 X        5.25 X
               [10.sup.0]  [10.sup.1]:1

Low

                   5.40 X        6.24 X
               [10.sup.0]  [10.sup.1]:1

Glaring conditions are bold.


The group of Fig. 7 I shows the mean values for each treatment for cognitive efficiency and effectiveness. The first part of the acronym stands for the task factor and its two levels, prosaccadic (P) and antisaccadic (A). The second letter is the size factor and its two levels: large (L) and small (S). The last part of the abbreviation is the luminance factor and its two levels: high (H) and low (L).

The left graphic illustrates similar mean values of cognitive efficiency (RTLX) by treatment. The right graphic shows the mean cognitive effectiveness (RST) by treatment, with lower WM span in high luminance treatments where vision was shifted. The statistical significance of these descriptive observations was tested.

4.1 COGNITIVE EFFECTIVENESS

The RST scale ranges from one to six, each level corresponding to the amount of words the subject can hold in memory. Interaction effects occur when there is a different response between the levels of a factor at different levels of other factors. The plots of interaction effects show the response of a dependent variable against the levels of an "A" factor for both levels of a "B" factor. Figure 8 (right) shows that when the source size was small and the task was antisaccade the RST performance was higher. Considering the same source sizes, working memory span was lower when the task was performed in a prosaccade way. In large source size treatments RST scores decreased more for prosaccade tasks, with no variations for antisaccade tasks.

[FIGURE 7 OMITTED]

[FIGURE 8 OMITTED]

For the combination of the task and luminance factors, there was a greater effect on the performance of RST at high luminance with a prosaccade task. When vision was shifted between the source and the computer screen, task performance was lower at high luminance contrasts. This observation provides evidence that the drop in performance was more related to visual adaptation than to eye movement itself. To support this conclusion, the literature indicates that in tasks involving visual-spatial working memory, the eye movement causes a decrease in performance [Theeuwes and others 2005, Smyth 1996]. This experiment, however, involved the executive part of WM because RST requires remembering and processing at the same time. As RST belongs to the central executive domain, eye movement effect in performance should be ruled out and the worse performance might be caused by light-related mechanisms. Finally, no interaction effect was found between size and luminance.

The main effects plots (Fig. 8 left) for this variable compare RST mean values (y-axis) for the lower and higher level of each factor (x-axis). The graphic shows better performance of RST in antisaccadic scenarios versus prosaccade treatments. The higher level of the luminance factor associated with a worse performance in RST, whereas the source size showed the same inverse effect: a lower working memory span was present at large source size scenarios.

A confirmatory analysis was carried out to verify the statistical significance of the observed effects. Table 2 shows the estimated effects on RST of every factor and their interaction. It also shows the results of an analysis of variance. Statistically significant relationships are highlighted.
TABLE 2. Estimated Effects and Analysis of Variance for RST.

                      Estimated Effects for RST
Term                           Effect             [rho]

Taskf (T)                                  0.230  0.019

Size (S)                                  -0.030  0.760

Luminance (L)                             -0.160  0.103

T*S                                        0.040  0.683

T*L                                        0.150  0.127

S*L                                        0.030  0.760

T*S*L                                     -0.080  0.415

                    Analysis of Variance for RST

Source                                         F  [rho]

Main effects                                2.76  0.042

2-way interactions                          0.87  0.458

3-way interactions                          0.67  0.415

Statistically significant relationships are bold.


[FIGURE 9 OMITTED]

The only factor that significantly affected performance on the RST was the way that the task was done: prosaccade or antisaccade (p-value=0.019). The cause of such variations may be the different attention management capabilities among subjects [Jarrold and Towse 2006]. The lighting source was a visual stressor that caused glare but also behaved as a cognitive stressor, reducing working memory capacity of the subjects and being a powerful environmental distractor. This effect might not be uniquely attributable to glare because it is confounded with the differing trial protocols between the prosaccadic and antisaccadic conditions.

Follows in order of importance were the effects caused by luminance and by the interaction between luminance and task. For this sample size and this experimental design, these effects were not enough to produce significant effects, yet they showed a trend.

4.2 COGNITIVE EFFICIENCY

The subjective scale used to assess cognitive efficiency of the volunteers during the experiment ranges from zero when there was no cognitive effort to a hundred when cognitive effort was the highest. Figure 9 illustrates the low effect of this experimental design factor on cognitive efficiency. For the combination of the factors task and size, when the size was small and the task antisaccadic, cognitive efficiency was lower than treatments where vision was shifted on the same source size. In large source treatments RTLX scores increased more under prosaccadic scenarios. An interaction effect between both factors was found. The variation detected, however, was close to 2 percent. For the combination of task and luminance factors, performing a prosaccadic RST resulted in a slightly greater cognitive effort. Finally, there is evidence of a small interaction effect between source size and luminance. When the source was small, an increase of luminance caused a decrease in RTLX level, but when the source was large, greater luminance increased RTLX score.

The main effects plot for this variable (Fig. 9) shows in its upper left, that when the RST was antisaccadic, cognitive efficiency was slightly smaller than when it was prosaccadic, alternating vision between the source and the task. Regarding the source size (upper right area of the graph) RTLX levels were higher al larger apparent sizes. However, the difference was around 2 percent. The difference of cognitive efficiency related to the source luminance was below 1 percent, so might be considered practically independent of that factor.

This expected result was found when the data of RTLX and RST were analyzed in the same order that they were obtained. When the first and the last treatment were compared, statistically significant variations of cognitive efficiency were found (p-value < 0.005) and no significant variations of cognitive effectiveness were found (p-value = 0.2539) confirming De Waard's model. We believe those results were due to fatigue effects. As the treatment order was randomized, the possible effects of fatigue were disarticulated when data was analyzed by treatment and RTLX failed to predict variations in cognitive efficiency.
TABLE 3. Estimated Effects and Analysis Variance for RST.

                      Estimated Effects for RTLX
Term                            Effect               [rho]

Taskf (T)                                  -0.801  0.671

Size (S)                                    1.009  0.593

Luminance (L)                               0.108  0.954

T*S                                        -0.508  0.788

T*L                                         0.075  0.968

S*L                                         0.517  0.784

T*S*L                                       1.683  0.373

                    Analysis of Variance for RTLX

Source                                          F  [rho]

Main effects                                 0.16  0.925

2-way interactions                           0.05  0.985

3-way interactions                           0.80  0.373

Statistically significant relationships are bold.


5 CONCLUSIONS

A new research line started with this study to determine the influence of lighting in cognitive performance. This paper tested the hypothesis that a large glare source produces adverse nonvisual effects in terms of cognitive efficiency and effectiveness on ICT operators. Cognitive efficiency was assessed by means of RTLX and cognitive effectiveness was measured as working memory span with RST. We built a large area light source emulating a window and a 2 x 2 x 2 full factorial experiment was designed, resulting in eight treatments. The factors were the luminance contrast between the task and the glare source, the apparent size of the glare source, and the way that the task was done, either shifting or not shifting vision between the glare source and the task.

The artificial window allowed us to explore the role lighting as a cognitive stressor, rather than the traditional approach as a visual stressor. To characterize lighting conditions on each treatment we built luminance maps from HDR images and then calculated DGI with Evalglare. We complemented objective glare analysis with the subjective sensation of glare of the participants with GSV scale.

The inclusion of the subjective variable RTLX was based on its greater sensitivity than performance objective methods. No statistically significant relationship between this variable and the factors selected was found. We conclude that RTLX was more sensitive to fatigue effects but failed to predict variations in cognitive efficiency caused by the glare source.

The results for the RST objective variable showed statistically significant effects of the task factor. In those treatments where the volunteers had to toggle their sight between the glare source and the task the cognitive effectiveness was lower. Lower RST scores were achieved in higher glare rated treatments. Considering those results we can state that the glare source has statistically significant effects on cognitive effectiveness when it briefly occupies the visual field. A caveat is that this effect might not be uniquely attributable to glare because it is confounded with the differing trial protocols employed for the prosaccadic and antisaccadic conditions.

Eye movement itself did not cause lower RST scores in prosaccadic treatments. Literature reports a correlation between eye movement and performance in visual-spatial working memory tasks [Theeuwes and others 2005, Smyth 1996] but the RST is under the central executive domain. To explain why a greater sense of glare was associated with lower working memory spans, we propose that the source acted as an attention distractor, linking working memory span with the attention management capabilities of the individuals [Jarrold and Towse 2006]. The source acted as a visual stressor causing glare, and also had effects as a cognitive stressor, reducing WM capacity of the subjects, behaving as a powerful environmental distractor. The proposed attention mechanism behind the changing glare conditions causing variations in working memory span has become a new hypothesis for further investigation. Finally, in order to avoid glare in real workspaces (and its associated visual and cognitive negative effects) we encourage the use of innovative strategies, new materials and combined natural and electric lighting design to control (and let the occupant control) the incoming light to create glare-free healthy environments.

REFERENCES

Arnetz BB, Wikholm C. 1997. Technological stress: Psychophysiological symptoms in modern offices. J Psychosom Res. 43(1):35-42.

Atkinson RC, Shiffrin RM. 1968. Human memory: A proposed system and its control processes. Psychol Learn Motiv. 2:89-195.

Baddeley AD, Hitch G. 1974. Working Memory. Psychol Learn Motiv. 8:47-904.

Baddeley AD. 2000. The episodic buffer: A new component of working memory. Trends Cogn Sci. 4(11):417-423.

Baron RA, Rea MS, Daniels SG. 1992. Effects of indoor lighting (illuminance and spectral distribution) on the performance of cognitive tasks and interpersonal behaviors: the potential mediating role of positive affect. Motiv Emotion. 16(1):1-33.

Bellia L, Cesarano A, Iuliano G, Spada G. 2008. Daylight glare: a review of discomfort indexes. Proceedings of visual quality and energy efficiency in indoor lighting: today for tomorrow. Rome (Italy). 19 p.

Biocca F. 2000. New media technology and youth: Trends in the evolution of new media. J Adolescent Health. 27(2):22-29.

Blehm C, Vishnu S, Khattak A, Mitra S, Yee R. 2005. Computer Vision Syndrome: A Review. Sury Ophthalmol. 50(3):253-262.

Boyce P, Hunter C, Howlett 0. 2003. The benefits of daylight through windows. Prepared under US Department of Energy Award No DE-FC26-02NT41497. 88 p.

Byers J, Bittner A, Hill S. 1989. Traditional and raw task load index (TLX) correlations: Are paired comparisons necessary? Adv Ind Erg I. 481-485.

Chauvel P, Collins JB, Dogniaux R, Longniore J. 1980. Glare from windows: Current views of the problem. Proceedings of Symposium on Daylight. Berlin (Germany). 9 p. [CIE] Commission International de l'Eclairage. 1987. International Lighting Vocabulary.

Geneve (Switzerland). CIE. Publication No. 17.4-1987. 379 p. [CIE] Commission International de l'Eclairage. 1995. Discomfort Glare in Interior Lighting. Vienna (Austria). CIE Technical Report No. 117-1995. 39 p.

Conway AR, Jarrold C, Kane MJ, Miyake A, Towse J. 2007. Variation in working memory: An introduction. In: Variation in Working Memory. Oxford (UK): Oxford University Press. 348 p.

Cowan N. 2010. The Magical Mystery Four: How Is Working Memory Capacity Limited and Why? Curr Dir Psychol Sci. 19(1):51-57.

Cuttle C. 2010. Towards the third stage of the lighting profession. Lighting Res Technol. 42(1):73-93.

Daneman M, Carpenter P. 1980. Individual differences in working memory and reading. J Verb Learn Verb Be. 19(4):450-466.

De Waard D. 1996. The Measurement of Drivers' Mental Workload. Traffic Research Centre. Groningen University (The Netherlands). PhD Thesis. 135 p.

Duff SC, Logie RH. 2001. Processing and storage in working memory span. Q J Exp Psychol. 54A(1):31-48.

Fonseca I, Porto MM, Fanchiotti A, Gonsalves A. 2006. Natural Light and Environmental Stress. In proceedings of 23rd PLEA. Geneva (Switzerland). 6 p.

Galasiu AD, Veitch JA. 2006. Occupant preferences and satisfaction with the luminous environment and control systems in daylit offices: A literature review. Energ Buildings. 38(7): 728-742.

Hallet PE. 1978. Primary and secondary saccades to goals defined by instructions. Vision Res. 18(10):1279-1296.

Hart S, Staveland L. 1988. Development of NASA-TUC (Task Load Index): Results of empirical and theoretical research. In: Human Mental Workload. Hancock PA, Meshkati N. Editors. Amsterdam (The Netherlands): Elsevier Science Publishing Co. 382 p.

Hashizume A, Kurosu M, Kaneko T. 2007. Multi-window system and the working memory. p. 297-305. In: Engineering psychology and cognitive ergonomics. Berlin (Germany): Springer Berlin / Heidelberg. .

Hoffmanna G, Guflera V, Griesmacherb A, Canazeic M, Stagglc S, Schobersbergera W. 2008. Effects of variable lighting intensities and colour temperatures on sulphatoxymelatonin and subjective mood in an experimental office workplace. Appl Ergon. 39:719-728.

Hopkinson RG, Bradley RC. 1960. A study of glare from very large sources. Illum Eng. 55(5): 288-294.

Hopkinson RG. 1972. Glare from daylighting in buildings. Appl Ergon. 3(1):206-215.

Hygge S, Knez I. 2001. Effects of noise, heat and indoor lighting on cognitive performance and self-reported affect. J Environ Psychol. 21:291-299.

Inanici M, Galvin J. 2004. Evaluation of high dynamic range photography as a luminance mapping technique. Lawrence Berkeley National Laboratory. LBNL Report No. 57545.28 p.

Iwata T, Kimura K, Shukuya M, Takano K. Discomfort caused by wide source glare. Energ Buildings. 199; 15(3-4):391-3981.

Iwata T, Shukuya M, Somekawa N. 1992. Experimental study on discomfort glare caused by windows. Subjective response to glare from simulated window. J Arch Planning Environ Eng. 432:21-31.

Jarrold C, Towse N. 2006. Individual differences in working memory. Neuroscience. 139(1): 39-50.

Kane MJ, Hambrick DZ, Wilheim SW, Payne 0, Engle RW. 2004. The generality of working memory capacity: A latent-variable approach to verbal and visuo-spatial memory span and reasoning. J Exp Psychol. 133(2):189-217.

Kim W, Ahn HT, Kim JT. 2008. A first approach to discomfort glare in the presence of non-uniform luminance. Build Environ. 43(11):1953-1960.

Kim W, Han H, Kim JT. 2009. The position index of a glare source at the borderline between comfort and discomfort (BCD) in the whole visual field. Build Environ. 44(5):1017-1023.

Knez I. 1995. Effects of indoor lighting on mood and cognition. J Environ Psychol. 15:39-51.

Kranzberg M, Pursell C. 1981. Historia de la tecnologia. La tecnica en occidente de la prehistoria a 1900. Barcelona: Gustavo Gilli.

Masuda Y. 1980. The information society and post-industrial society. Washington (DC): World Future Society. 172 p.

Miller GA. 1956. The magical number seven, plus or minus two -some limits on our capacity for processing information. Psychol Rev. 63(1):81-97.

Miyake A, Shah P. 1999. Models of working memory: Mechanisms of active maintenance and executive control. New York (NY): Cambridge University Press. 506 p.

NUTEK. 1994. Office lighting. 34 examples of good and energy efficient office lighting. Stockholm (Sweden). Department of energy efficiency.

O'Donnell R, Eggemeier F. 1986. Chapter 42: Workload assessment methodology. In: Handbook of perception and human performance II. New York (NY). John Wiley and Sons. 1,358 p.

Osterhaus W, Bailey I. 1992. Large area glare sources and their effect on discomfort and visual performance at computer workstations. Proceedings of IEEE Industry Applications Society Annual Meeting. Houston (TX). 8 p.

Osterhaus W. 1993. Office lighting: A review of 80 years of standards and recommendations. Proceedings of the IEEE Industry Applications Society Annual Meeting. Toronto (Canada). 11 p.

Osterhaus W. 2005. Discomfort glare assessment and prevention for daylight applications in office environments. Sol Ener 79:140-158.

Repovs A, Baddeley A. 2006. The multi-component model of working memory: explorations in experimental cognitive psychology. Neuroscience. 139:5-21.

Roberta M, Savoiea N. 2006. Are there gender differences in verbal and visuospatial working-memory resources? Eur J Cogn Psychol. 18(3):378-397.

Rodriguez R, Pattini A. 2011. An online tool to identify white-collar worker profiles in relation to their ICT skills and mental strain. Cog Tech Work. 13(2):81-91.

Rubio S, Diaz E, Martin J, Puente JM. 2004. Evaluation of Subjective Mental Workload: A Comparison of SWAT, NASA-TLX and Workload Profile Methods. Appl Psychol-Int Rev. 53(1): 61-86.

Smyth MM. 1996. Interference with rehearsal in spatial working memory in the absence of eye movements. Q J Exp Psychol A. 49(4):940-949.

Theeuwes J, Olivers CNL, Chizk CL. 2005. Remembering a location makes the eyes curve away. Psychol Sci. 16:196-199.

Tokura M, Iwata T, Shukuya M. 1996. Experimental study on discomfort glare caused by windows, part 3. Development of a method for evaluating discomfort glare from a large light source. Journal of Architecture, Planning and Environmental Engineering. 489:17-25.

Tuaycharoen N, Tregenza PR. 2007. View and discomfort glare from windows. Lighting Res Technol. 39(2):185-200.

Unsworth N, Schrock J, Engle RW. 2004. Working Memory Capacity and the AntisaccadeTask: Individual Differences in Voluntary Saccade Control. J Exp Psychol Learn. 30(6):1302-1321.

Van Bommel W. 2006. Non-visual biological effect of lighting and the practical meaning for lighting for work. Appl Ergon. 37(4):461-466.

Veitch J, Newsham G. 1996. Determinants of lighting quality I: State of the science. Proceedings of Annual Conference of the illuminating Engineering Society of North America. Cleveland (OH).

Ward Larson G, Shakespeare R. 1998. Rendering with Radiance. San Francisco (CA): Morgan Kaufmann. 664 p.

Wastlund E. 2007. Experimental Studies of Human-Computer Interaction: Working memory and mental workload in complex cognition. Department of Psychology. Gothenburg University (Sweden). PhD Thesis. 40 p.

(1.) Laboratorio de Ambiente Humano y Vivienda. INCIHUSA - CONICET, Argentina

* Corresponding author: Roberto G. Rodriguez, E-mail: rgrodriguez@mendoza-conicet.gov.ar

Roberto G. Rodriguez PhD (1*) and Andrea Pattini PhD (1)

doi: 10.1582/LEUKOS.2012.08.04.003
COPYRIGHT 2012 Illuminating Engineering Society
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2012 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Rodriguez, Roberto G.; Pattini, Andrea
Publication:Leukos
Article Type:Report
Geographic Code:1USA
Date:Apr 1, 2012
Words:7725
Previous Article:Comparing judgments of visual clarity and spatial brightness through an analysis of studies using the category rating procedure.
Next Article:Design principle study of high efficiency compact fluorescent lamps.
Topics:

Terms of use | Privacy policy | Copyright © 2021 Farlex, Inc. | Feedback | For webmasters |