425 lines
28 KiB
Plaintext
425 lines
28 KiB
Plaintext
The Act of Task Difficulty and Eye-movement Frequency for the 'Oculo-motor indices'
|
|
|
|
Minoru Nakayama
|
|
|
|
Koji Takahashi
|
|
|
|
CRADLE, Tokyo Institute of Technology
|
|
|
|
CRADLE, Tokyo Institute of Technology
|
|
|
|
nakayarna@ cradle, titech.ac.jp
|
|
|
|
Yasutaka Shimizu
|
|
|
|
National Institute of Educational Policy Research
|
|
|
|
yasu@nier.go.jp
|
|
|
|
Abstract
|
|
|
|
The oculo-motor re ects the viewer s ability to process visual information. This paper examines whether the oculo-motor was affected by two factors: rstly task dif culty and secondly eye-movement frequency. In this paper, oculo-motor indices were de ned as measurements of pupil size, blink and eye-movement_ For the purpose of this study, two experiments were designed based on previous subsequcntial ocular tasks were subjects were required to solve a series of mathematical problems and to orally report their calculations.
|
|
The results of this experiment found that pupil size and blink rate increased in response to task dif culty in the oral calculation group. In contrast however both the saccade occurrence rate and saccade length were found to decrease with the increased difculty of the task. The results suggests that oculo-motor indices respond to task dif culty. Secondly, eye-movement frequencies were elicited by the switching frequency of a visual target. Pupil size and the saccade time were found to increase with the frequency however, blink and gazing time were found to decrease in response to the frequency. There was a negative correlation between blinking and gazing time. Additionally, the correlation between blinking and saccade time appeared in the higher frequencies.
|
|
These results indicate the oculo-motor indices are affected by both task dif eulty and eye-movement frequency. Furthermore, eye-movement frequency appears to play a different role than that of task dif culty.
|
|
|
|
CR Categories:
|
|
|
|
H.1.2 [User/Machine Systems]: Hu-
|
|
|
|
man information processing [H.5.2]: User Interfaces
|
|
|
|
Evaluation/methodology
|
|
|
|
Keywords: Eye-movement, Pupil Size, Blink, Saccade, Gaze
|
|
|
|
1 Introduction
|
|
The oculo-motor systems are driven by a viewer s mental activity such as problem solving. It has been found that eye pupil
|
|
|
|
and blink respond to psychological and cognitive dif culties[Beatty 1982][Tada et al. 1991]. It has also been noted that pupil size increases in accordance to increased task difculty [Beatty 1982] and emotions[Hess and Polt 1964]. Therefore this information can be used as an index to estimate an individual s Mental Workload (MWL) [ISO 1991] for a task[Osuga 1992].
|
|
Eye-movement is also known to change in accordance to the level of mental activity an individual is engaging in[Takeda 1976]. The change of saccade and gaze happens in response to the change of blink or pupil size[Takahashi et al_ 2000b], however what still remains unclear is the role and possible relationship between eyemovement and task difculty. An issue of particular importance was the attempt to establish which particular components of eyemovement were signi candy affected by a change in task dif culty. In this paper, some measures were compared in order to extract possible relationships, these evaluations were then used to construct oculo-motor indices that acted as measurements for pupil size, blink, saccade and gaze. The oculo-motor indices changed differently among the viewing tasks of our previous experiments: simple subsequential ocular task with an oral response task, a driving simulation task which was conducted with a video-game computer[Takahashi et al. 2000a][Takahashi et al. 2000c][Takahashi et al. 2000b]. It was hypothesized that a viewer s mental workload would increase in response to the oral response task and that a similar phenomena would occur in the driving-simulation condition. However, it was assumed that the responses of saccade, blink and gaze would be different amongst the various tasks. Some factors were considered as reasons for the possible differences between the two experimental conditions: for example subject s attention, monotony, viewing eld size, etc. There was also a possibility that the frequency of eye-movement could be a contributing factor in the observed differences.
|
|
The aim of this paper was to examine whether both task dif culty and eye-movement frequency affect oculo-motor indices and the relationship among them. The following issues were addressed:
|
|
To examine the changes in eye-movement in response to an increase in a viewer s mental activity during a subsequential ocular task.
|
|
|
|
Pen-nieslon to make digital or hard copies of all or part of this work for personal or classroom use is gmntad, without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this noUco and the full citaUon on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires pdor specific perrnlssion and/or a [Be. ETRA'02 New Odeans Louisiana USA Copyright ACM 20022 1-58113-447-3/02/03...Ik5.00
|
|
|
|
2. To examine the relationship among blink, saccade - - gaze in response to the frequency of eye-movement with oral questions.
|
|
For these purposes, two experiments were devised and conducted based upon subsequential ocular task with oral questions.
|
|
|
|
37
|
|
|
|
radiu~ of vi~unal 'dinnuli(5(h'~)
|
|
|
|
t ...... /
|
|
|
|
Igl
|
|
showing target
|
|
II
|
|
appearance position of the target
|
|
|
|
target interval=l see
|
|
|
|
Figure 1: The subsequential ocular task
|
|
2 Experiment I
|
|
This experiment was conducted in order to examine whether eyemovement changed in response to task dif culty during a subsequential ocular task.
|
|
2.1 Experimental procedure
|
|
2.1.1 Stimuli
|
|
Subsequential eye-movement were induced in order to simulate natural viewing through the use of a visual stimuli. Subjects were asked to follow a target symbol (+) which switched periodically in each second on the screen. The sample of the stimuli is illustrated in Figure 1. The symbol switched randomly on a circle whose radius was 5 deg in visual angle. Every visual angle between the two sequential target positions was xed to the radius at 5 deg. The visual angle of gazing was suggested at around 3 deg[Inui 1996], therefore it was assumed that every stimuli switch would motivate saccadic eye-movement. This visual stimuli was displayed on a 37 inch video monitor. A pair of speakers were placed on both sides of the monitor for providing audio instruction. The audio instructions were pre-recorded to a Mini disc, and then played back to the viewer at the same sound level.
|
|
2.1.2 Oral calculation task
|
|
The calculation tasks were given to subjects in order to activate mental activity during the subsequential ocular task. The audio instruction for an oral calculation was provided as a stimuli. The task consisted of one or two digits multiplication. For example, 23 x 6, 8 x 6, etc. Subjects were required to solve the multiplication problem and to give an oral answer of calculation process[Takahashi et al. 2000b]_
|
|
2.1.3 Oculo-motors
|
|
The eye-movement of the left eye was observed by an eye-tracker (NAC:EMR-7), which was synchronized with pupil observation. The eye-tracker output came in the form of x-y coordinates in i 0.1deg spatial resolution which signify the degree of eye tracking during the task. The sampling rate was 30Hz and the sessions were recorded on NTSC video.
|
|
The eye-movement speed was evaluated as differential angles per NTSC frame. To extract the saccade, the eye-movement was divided into saccade and gaze by the speed 40deg/sec[Ebisawa and Sugiura 1998]. In addition, the saccade length was computed by the summation of differential x-y coordinates of a saecade[Takahashi et al. 2000b].
|
|
|
|
1.2
|
|
|
|
o "~
|
|
|
|
I.I-
|
|
|
|
~'~1 . 0 ~
|
|
|
|
0.9
|
|
|
|
""1----
|
|
|
|
--"1""-
|
|
|
|
r--
|
|
|
|
control
|
|
|
|
correct
|
|
|
|
correct incorrect
|
|
|
|
[fast]
|
|
|
|
[slow]
|
|
|
|
Figure 2: Pupillary change with the task dif culty
|
|
|
|
1.0
|
|
|
|
~-.0.8-
|
|
,q
|
|
|
|
0.6-
|
|
v
|
|
0.4-
|
|
|
|
f
|
|
|
|
m 0.2-
|
|
|
|
0.0
|
|
|
|
I
|
|
|
|
I
|
|
|
|
I
|
|
|
|
control
|
|
|
|
correct
|
|
|
|
correct incorrect
|
|
|
|
[fast]
|
|
|
|
[slow]
|
|
|
|
Figure 3: Blink rate change with the task dif eulty
|
|
|
|
Pupil size was measured by processing the image of the eye supplied by the tracker. Blinking was detected and de ned as those period where the pupil appeared to be smaller. As blinking prevented eye-movement from being detected, the data for eye-movement during the blink were not included. Both pupil and blink were also observed at 30Hz.
|
|
2.1.4 Design
|
|
Seven male university students took part in this experiment as subjects and were paid for their time. Subjects were required to respond to ten tasks, each within 10see intervals. The following stimuli session was also conducted without any instruction as a control session.
|
|
2.2 Results
|
|
2.2.1 Definition of task difficulty
|
|
Subject s responses were divided into four conditions: control without calculation task (control), correct quick rresponsc (fast correct), correct slow response (slow correct) and incorrect response (incorrec0. The average percentage correct was 80.8%, and correct responses were divided into two groups based upon subjects reaction times. Task dif culty seemed to increase response times and this
|
|
|
|
38
|
|
|
|
1.2
|
|
|
|
~-~1.0-
|
|
|
|
• 0.8-
|
|
g
|
|
|
|
~0.6
|
|
|
|
I
|
|
|
|
I
|
|
|
|
I
|
|
|
|
control
|
|
|
|
correct
|
|
|
|
correct incorrect
|
|
|
|
[fast]
|
|
|
|
[slow]
|
|
|
|
Figure 4: Saccade occurrence rate with the task difculty
|
|
|
|
m
|
|
,~4-
|
|
t
|
|
d
|
|
|
|
1
|
|
control correct
|
|
[fast]
|
|
|
|
F correct
|
|
[slow]
|
|
|
|
I
|
|
incorrect
|
|
|
|
Figure 5: S accade length with the task dif culty
|
|
|
|
was found to be the same for all four conditions. Most subjects also supported this order as the task dif culty.
|
|
Pupil size was compared among the four response conditions. Figure 2 shows average pupil size with standard error across four conditions. Pupil size was found to increase from control to incorrect response. An analysis of variance (ANOVA) was performed on the data in order to ascertain if pupil size could be affected by task dif culty. The source of response conditions was signi cant (F(3,295)=18.5, p < 0.01) when performing a one-way ANOVA. The pupil size for the response session was signi candy larger than for the control session (p < 0.01). As with previous studies this results suggests that pupil size re ects task dif culty[Beatty 1982].
|
|
Average blink rate with standard error was also summarized in Figure 3. The gure shows that the blink rate increases in accordance to response conditions. The source of the condition found to be signi cant for the blink rate when performing a oneway ANOVA (F(3,363)=9.4, p < 0.01). The blink rates for correct[slow] and incorrect responses were signi cantly higher than those for the control session (p < 0.05). In particular, the blink rate for incorrect responses were twice those found for the control session. This nding suggests that subjects blinked more frequently when issuing incorrect responses. Both the results of Figure 2 and Figure 3 concur with subject s impression and previous studies[Beatty 1982][Tada et al. 1991 ]. These ndings suggest that task dif culty can be de ned by both pupil and blink changes.
|
|
2.2.2 Thesaccade occurrence rate
|
|
To examine the in uence of blink to the saccade, the saccade frequency for the response conditions were compared, because of the noted negative correlation between the blink rate and the saccade rate as suggested in a previous study[Takahashi et al. 2000b].
|
|
The occurrence rate of saecade was de ned as the number of saccades per second. The change in the occurrence rate of saccade with standard error was illuslrated in Figure 4. The rate for the control session was around 1.0 saccade/sec when responding to the stimuli switching interval. The rate for the correct[fast] was the same as that for the control session, however the rate decreased in the response condition of correct[slow] and incorrect. The source of the response condition was found to be signi cant when analysing the results using a one-way ANOVA (F(3,241)=10.5, p < 0.01). The rate for incorrect response was signi candy lower than the control and fast correct response (p < 0.05).
|
|
|
|
2.2.3 The saccade length
|
|
In order to compensate for the reduced saceade frequency, the length of saccade had to be controlled. The average length of the saccade was then compared across the response conditions. The average length of saccade with standard error was summarized across the conditions in Figure 5. The gure shows that the length was shortened within the response conditions. To examine the change of length, an ANOVA was conducted. The source of the response condition was also signi cant (F(3,242)=6.0, p < 0.01). The results suggest that the length for an incorrect response was signi cantly shorter than for the control and fast correct response.
|
|
2.3 Discussion
|
|
Total gazing time, another measure of eye-movement, was found to signi candy decrease across the response conditions. Conducting ANOVA, the source of the condition was signi cant (_F(3,248) = 12.3, p < 0.01). The relative time for gazing was compared across the conditions and results revealed that the average time for both slow correct and incorrect response was signi cantly lower than that for the control.
|
|
There was no signi cant difference for all indices between the control condition and fast correct response without pupil size. Despite gazing time signi cantly decreasing in the group of incorrect responses, for the incorrect response the saccade occurrence rate was the lowest and more signi cantly the saccade length was the shortest. This suggests that eye-movement measures are in uenced by the increase of invisible time when blink rates are higher. As a result, eye-movement suppression appears in response to mental task difculty. Furthermore, it is very interesting that task dif culty affects eye-movement behavior despite subjects having enough time in slow subsequential ocular tasks. This lading provides further evidence that task dif culty in uences measures of eye-movement_
|
|
3 Experiment II
|
|
This experiment was conducted in order to examine whether the relationship among oculo-motor indices would be signi cantly affected by eye-movement frequency.
|
|
|
|
39
|
|
|
|
1.2
|
|
|
|
• ]1-
|
|
|
|
;>
|
|
c~ 1.0-
|
|
|
|
0.9
|
|
0.0
|
|
|
|
A incorrect ---e--- correct
|
|
|
|
--II-- control
|
|
|
|
I
|
|
|
|
I
|
|
|
|
I
|
|
|
|
I
|
|
|
|
I
|
|
|
|
0.5 1.0 1.5 2.0 2.5 3.0
|
|
|
|
Switching frequency (Hz)
|
|
|
|
Figure 6: Relative pupil size with the target switching frequency
|
|
3.1 Experimental procedure
|
|
The same subsequential task in Experiment I was conducted. Eyemovement frequency was controlled in all ve levels by switching the frequency of the target symbol. An oral response task was also administered to subjects and was the same as that noted in Experiment I.
|
|
Four male university students aged between 22 and 25 years of age took part in this experiment as subjects and were paid for their time. Switching frequencies of the target were controlled at 0.8, 1.0, 1.5, 2.0 and 2.5Hz. The switching target could evoke eyemovement in the same frequency. The control session was also conducted without any audio instruction. Therefore, the target switching frequency was added to the experimental parameters as laid out in Experiment I.
|
|
3.2 Results
|
|
In Experiment II, the responses were divided into three conditions: control, correct, and incorrect response. Because there was little difference between fast correct and slow correct in Experiment I. The percentage correct for the oral task was 83%. The eyemovements were frequent in response to the target switching, blink, saecade and gaze were evaluated as costing time rate[Takahashi et al. 2000b] in Experiment 1I.
|
|
3.2.1 Pupil response
|
|
Pupil size was compared among the three response conditions along with the target switching frequency. Figure 6 shows the change of the average pupil size with standard error. Pupil size increased with the response condition in the same order as that found in Experiment I. In addition, pupil size increased with target switching frequency. An ANOVA was conducted to assess the signi cance of the pupillary change, both sources of the response condition and the target switching frequency were found to be signi cant (the response condition: F(2,492)=19.2, p < 0.01; the target switching frequency: F(4,492)=19.9, p < 0.01), and the interaction of those sources was not however signi cant (F(8,492)=0.6, p = 0.75). This
|
|
nding suggests that an increased switching frequency increases task dif culty and mental workload thereby making it harder to follow the visual target.
|
|
|
|
14
|
|
12-
|
|
~. 10-
|
|
~ 8-
|
|
E
|
|
"~ 6-
|
|
. ....q
|
|
~ 4-
|
|
_
|
|
0
|
|
0.0
|
|
|
|
A incorrect •- e - - c o r r e c t ---II--- control
|
|
|
|
I
|
|
|
|
I
|
|
|
|
I
|
|
|
|
I
|
|
|
|
I
|
|
|
|
0.5 1.0 1.5 2.0 2.5 3.0
|
|
|
|
Switching frequency (Hz)
|
|
|
|
Figure 7: Blink time rate with the target switching frequency
|
|
|
|
3.2.2 Blink response
|
|
The change of blinking time rate was summarized using the same format in Figure 7. This result suggests that blink costs time for 6 11% of the total observed time. Blinking times were seen to increase across the response conditions as well as in Experiment I. However, blink decreases were noted with target switching frequency. The differences of blinks can be suppressed according to the target switching frequency. If this measure is taken then there is a small differences between the correct response and the control in higher frequency condition of target switching. Both sources of the task dif culty and target switching frequency were analysed using a two-way ANOVA and were found to be signi cant (response condition: F(2,371)=5.7, p <7 0.01; switching frequency: F(4,371)=9.3, p < 0.01), the interaction of those sources was not signi cant (F(8,371)=0.2, p = 0.98).
|
|
In Experiment I, both pupil size and blink rate were found to increase with in line with an increase in task dif culty. Both were also seen to increase in line with task dif culty in Experiment II. Pupil size and blink do however reveal a different tendency with the target switching frequency measure. This result suggests that blink is suppressed by frequent eye-movements which are caused by the following of a visual target.
|
|
3.2.3 Gazing time
|
|
A s blink time decreases, gazing time should change in accordance to itsrelationship to blink time. The gazing time, which was de ned as the rateof gazing for the totalobservation period, are illustrated in Figure g. The gazing time rate decreased across the response conditions as well as in Experiment I.
|
|
A deviation of gazing time for the control session was found to be more frequent than that for incorrect responses. The deviation decreased for higher frequencies of target switching. Both sources of task dif culty and the target switching frequency were found to be signi cant during an A N O V A analysis (response condition: F(2,394)=3.0, p <7 0.05; switching frequency: F(4,394)=4.2, p <7 0.01). The interaction of the factors was not signi cant (F(8,394)=0.3, p = 0.96). T h e result of this A N O V A suggests that gazing time rates decrease with task dif culty and that of target switching frequency.
|
|
|
|
40
|
|
|
|
85
|
|
O
|
|
I~ 80-
|
|
..--4 tN
|
|
|
|
--B-- control
|
|
|
|
+
|
|
|
|
correct
|
|
|
|
,t incorrect
|
|
|
|
75
|
|
|
|
t
|
|
|
|
I
|
|
|
|
I
|
|
|
|
I
|
|
|
|
I
|
|
|
|
0.0 0.5 1.0 1.5 2.0 2.5 3.0
|
|
|
|
Switching frequency (Hz)
|
|
|
|
Figure 8: Gazing time rate with the target switching frequency
|
|
|
|
Table 1: Correlation coef cient between blinking and gazing time
|
|
|
|
target switching frequency (Hz) blinking time ---gazing time
|
|
|
|
0.8
|
|
|
|
1.0
|
|
|
|
1.5
|
|
|
|
2.0
|
|
|
|
-0.97 -0.97 -0.95 -0.90 co<o..]) co<o..J) (p<o.oo ~<o.oo
|
|
|
|
2.5
|
|
-0.84 (p<o.o,)
|
|
|
|
25
|
|
|
|
,_,20
|
|
|
|
15-
|
|
|
|
o lO-
|
|
0
|
|
5-
|
|
0
|
|
0.0
|
|
|
|
---II-- control ---e- correct
|
|
A incorrect
|
|
|
|
I
|
|
|
|
I
|
|
|
|
I
|
|
|
|
I
|
|
|
|
I
|
|
|
|
0.5 1.0 1.5 2.0 2.5 3.0
|
|
|
|
Switching frequency (Hz)
|
|
|
|
Figure 9: Saccade time rate with the target switching frequency
|
|
|
|
Table 2: Correlation coef cient between blinking and saccade time
|
|
|
|
target switching frequency (Hz) blinking time ---~accade time
|
|
|
|
0.8
|
|
|
|
1.0
|
|
|
|
1.5
|
|
|
|
2.0
|
|
|
|
2.5
|
|
|
|
-0.04 -0.21 -0.23 -0.37 -0.43 ~=0.75) (p<o.]n) (p<o_os) (p<o.o,) ~<o.o])
|
|
|
|
3.2.4 Correlation of blinking and gazing time
|
|
A trade-off relationship is suggested between blinking and gazing time in problem solving situations[Takahashi et al. 2000b]. The possibility of relationships within this study s data set were also examined. Correlation cocf cients are summarized in Table 1.
|
|
All correlation coef cients were found to be signi cant and showed a strong negative correlation. The trade-off relationship was examined within those coef cients. It was found that blinking time rates increased within the response condition, despite gazing time rates decreasing within the condition. Both these changes consisted of negative correlations.
|
|
Additionally, the absolute value of the coef cient slightly decreased with the target switching frequency (-0.97 to -0.84). The variance of both time rates were suppressed according to the switching frequency. It was considered that the decrease of variance affects to change for the absolute value of the coef cient.
|
|
3.2.5 Saccade time
|
|
The target switching frequency must affect the saccade directly. To examine the effect of the frequency to the saccade, the saecade time rates with standard error are summarized in Figure 9. Figure shows that the saccade time rate monotonically increases with frequency. The deviation of the saccade time among the response conditions also increases with the frequency. Two-way ANOVA was conducted on the saccade time rate. Both sources of the task dif culty and the target switching frequency were found signi cant (response condition: F(2,520)=5.8, p < 0.01; switching frequency: F(4,520)=42.8, p < 0.01). The interaction of both sources was not signi cant (F(8,520)=0.8, p = 0.58). The result suggests that both task dif culty and switching frequency affect saccade time signi cantly, and similarly task dif culty reduces the saccade time to
|
|
|
|
follow the target.
|
|
3.2.6 Correlation of blinking and gazing time
|
|
A viewer has to have some invisible time during both blinking and saccade time. The trade-off relationship has been suggested between both blinking and saccade time[Takahashi et al. 2000b]. The correlation coef cients are summarized in the same format as in Table 2.
|
|
Results show that the correlation coef cient between blink time rate and saccade time rate gradually changc.d according to the target switching frequencies (r=-0.04 to -0.43). There was little correlation in the area of lower frequencies of eye-movement, but a signi cant correlation appears in frequent eye-movements over 1.5-2.0Hz. 'ntis suggests that the trade-off relationship is dependent on eye-movement frequency. Because the eye needs to keep visible time despite invisible time increasing such as the periods when the frequeneyofeye-movementishigber, thena trade-off relationship appears between blink and saccade occurrence.
|
|
3.3 Discussion
|
|
Experiment 1I examines whether the factor of eye-movement frequency affects the oculo-motors. The target switching frequency may evoke the eye-movement frequency. Increasing the frequency of a visual target to follow may lead to a viewer s mental workload to be increased.
|
|
3.3.1 The frequency factor
|
|
Pupil size increases within the response conditions as task dif culty increases. This increase is also seen to occur when the target
|
|
|
|
41
|
|
|
|
switching frequency condition is applied. Both increases indicate that the viewer s mental workload increase and that this increased workload is an important factor in experimental situations. Blinking time also increases with the task dif culty of an oral calculation. This response was found to occur in Experiment I. However, blinking time decreases with the target switching frequency. The responses show different and opposite consequences between these two factors of dif culty and of frequency. This result suggests that task dif culty and eye-movement frequency are different factors of change of oculo-motor indices .
|
|
The decrease of gazing time with task dif culty as noted in Experiment I was a nding in line with a previous study[Takahashi et al. 2000b]. Gazing time was also seen to decrease with target switching frequency. Therefore, both task dif culty and eyemovement frequency can be said to affect the decrease found in gazing time. The saccade decreases with task dif culty and was noted in the ndings of Experiment I and a previous study[Takahashi et al. 2000b]. According to the target switching frequency, saccade time increases monotonically. As a result, both factors of task dif culty and eye-movement frequency affect oculo-motor indices . Additionally, a change of the indices for task dif culty is independent on eye-movement frequency.
|
|
3.3.2 'Trade-off' relationship
|
|
The correlation coef cient in Table 1 indicates a strong negative correlation between blink and gazing time. The absolute value of the correlation coef cient changes in a smaller range with the frequency. This trade-off relationship has also been shown in the previous studies[Takahashi et al. 2000a][Takahashi et al. 2000b]. However, the responses of blink and gaze were different among those. In this experiment, both task dif culty and eye-movement frequency were found to affect blink and gaze respectively. Therefore, it is interesting to consider which factor mainly affects the negative correlation. According to Figure 7 and 8, both deviation of the change are suppressed in response to the target switching frequency. That suppression may affect the slight decrease of the coef cient with the frequency.
|
|
The factor of another Irade-off relationship between blink and saccade time still remains unclear, as this relation appears to depend on the particulars of the experimental condition. Table 2 shows a change in the correlation coef cient between blink and saccade time. For lower frequencies of eye-movement, the correlation coef cient is not signi cant. This suggests that task dif culty bears little in uence to this relationship, as task dif culty is independent of the frequency. However, the relationship appears to be stronger when eye-movement frequency is higher. Therefore, the trade-off relationship between the blink and saccade time depends on higher eye-movement frequency.
|
|
4 Conclusions
|
|
This study was constructed to critically examine evaluate the oculomotor and the role of eye-movement changes, task dif culty and mental workload. Two experiments based on subsequential ocular tasks with oral calculation were conducted.
|
|
Firstly, both pupil size and blink rates were found to increase with task dif culty. Additionally both the saccade occurrence rate and the saccade length decreased in response to task dif culty. These results suggest that oculo-motor indices change according to task dif culty.
|
|
Secondly, eye-movement frequency appears to affect the oculomotors. It was found that the responses to the frequency are different for task dif cul ty, and that eye-movement frequency plays a role as another factor for oculo-motor indices . The trade-off relation
|
|
|
|
between the blink and saccade time depends on eye-movement frequency.
|
|
The results provide evidence that the oculo-motors can be a useful index for task dif culty and that eye-movement frequency affects the measures of the oculo-motor. To inspect how both factors of task dif culty and eye-movement operate in natural viewing conditions and to extract other factors of oculo-motor change are issues that require further research attention_
|
|
5 Acknowledgement
|
|
This research was partially supported by the Japanese Ministry of Education, Sports, Culture, Science and Technology(MEXT), Grant-in-Aid for Encouragement of Young Scientists, 0380, 1998-2000 and Grand-in-Aid for Scienti c research, (B)(2)--1-0558020, 1998--2000, (C)(2)--!~3680228, 2001-kO02.
|
|
References
|
|
BEATTY, J_ 1982. Task-evoked pupillary resposes,processing load, and the structure of processing resources. Phychological Bulletin 91, 2, 276--,392.
|
|
EBISAWA, Y_, AND SUGIURA, M. 1998. In uences of target and
|
|
xation point conditions on characteristics of visually guided voluntary saccade. The Journal of the Institute of Image Information and Television Engineers 52, 11, 1730--1,737.
|
|
HESS, E_ H., AND POLT, J . M . 1964. Pupil size in relation to mental activity during simple problem solving. Science, 132,
|
|
1190--1-192.
|
|
INUI, T. 1996. Brain and visual congnition (in Japanese). Science sha.
|
|
ISO. 1991. Ergonomic principles related to mental work-load ---General terms and definitions. International Organization for Standardization.
|
|
OSUGA, M. 1992. Evaluation of mental workload using physiological responses. Journal of the Society of lnstrument and Control Engineers 31, 2, 324 325.
|
|
TADA, H., YAMADA, F., AND FUKUDA, K. 1991. Psychological blink (in Japanese). Kita Ouji Shobo.
|
|
TAKAHASHI, K., N A K A Y A M A , M., A N D SHIMIZU, Y. 2000. Eyemovements and pupillarychanges dunng simulated driving.The Journal of the Institute of Image Information and Television Engineers 54, 9, 1323-4329.
|
|
TAKAHASHI, K., NAKAYAMA, M., AND SHIMIZU, Y. 2000. The response of eye-movement and pupil size to audio instruction while viewing a moving target. In Eye Tracking Research and Applications Symposium 2000, A C M Press, New York, S. N. Spencer, Ed., ACM, 131--438.
|
|
TAKAHASHI, K_, N A K A Y A M A , M., A N D SHIMIZU, Y. 2000. Taskevoked cye-movernent and pupilresponses. Perception, Supplement 29, 36.
|
|
TAKEDA, M. 1976. Effect of mental activity in problem solving on eye movements_ The Japanese Journal of Ergonomics 12, 5, 175--t-81.
|
|
|
|
42
|
|
|