HUMAN
F A C T O R S , 1979,21(3),331-336
Information Transmission during Eye Movements ALBERT E. BARTZ', Departriiciit of Psycliolog): Coiicordia College, hfoorliead, Afiiiiiesota
Target stiiiiiili (the iiiiriieral 5 at 60 degrees to the left of the siibject's fixatioii point) i w - e extirigiiislied at varioiis tiiiies dirririg the siibject's eTe iiioiviiieiit to that stiiiiiiliis aiid the subject was asked to giiess what iiirrtieral had occiirred. Tlircshold frirrctioris were coiistnrcted from tliese data, arid it 1 1 ~foiriid s that at the begiririirig of the backward corripisatoty iiioveriierit, tary little iiifoniiatioii rt*astrarisriiitted. Hoit*e\vr,as tlie backi\wrd coiiipeiisatoty riioiwiieiit progressed, there litas ari iiicrease iri the correct ideiitificatioii of tlic stiriiirliis, arid a tliresliokd it*as reached before die fom*ardcoriipeiisatory iiiotwierit begati.
INTRODUCTION During the last decade there has been a renewed interest in the role of eye movements in visual perception due, in part, to the development of new techniques and the refinement of older methods for more precise eye movement recording. Research studies have ranged from descriptions of the physiology of eye movement control (e.g., Fuchs and Luschei, 1970; Skavenski and Robinson, 1973; Shebilske, 1976) to the role of eye movements in "higher" mental functions such as search behavior (Could, 1973; Mackworth, 1976). reading (Spragins, Lefton, and Fisher, 1976), and picture scanning and recognition (Noton and Stark, 1971; Antes, 1974). Basic to understanding the contribution of eye movements to a complex visual pcrceptual task is determining the manner in which information is processed during and after the eye movement itself. Several studies have investigated the change in the sensitivity of the Requests for reprints should be sent to Prof. Albert E. Bartz, Department of Psychology, Concordia College, hloorhead, hlinnesota 56560. U S A .
eye just prior to and during a saccadic movement (e.g.. Latour, 1962; Zuber and Stark, 1966; Volkmann, Schick, and Riggs, 1968; Ritter, 1976). In studies.of this type a short flash which is just above threshold (probability of detection = 1.0) for the stationary eye is presented at various times just prior to, during, and just after the voluntary eye movement. The subject then reports whether or not he saw the flash, and a large number of trials are run to determine the threshold from the detection function. Typical results usually show a saccadic siippressioii in that the detection threshold drops from 100% about 100 ms prior to the saccade to near 0% shortly after the saccade begins, and climbs gradually to 100% about 150 ms after the completion of the saccade (Volkmann, Schick, and Riggs, 1968; Volkmann, 1976).
Although the above studies have contributed greatly to the concept of visual sensitivity during eye movement, the detection of a threshold-level test flash represents a rather primitive level of functioning. In addition, the type of task needed to obtain the data re-
@ 1979, The Human Factors Society, Inc. All rights reserved.
33 1
332-June, 1979
quires the subject to make relatively restricted responses in that no head movements a r e permitted and the visual field to be searched is greatly reduced. When the subject is permitted to make a more “natural” response, i.e., he is permitted to move both head and eyes freely to fixate a peripheral stimulus that is more than 20 degrees from central fixation, the eye movement becomes more complex (Robinson, Koth, and Ringcnbach, 1976). It has been shown that at the onset of a randomly presented peripheral stimulus, the eyes begin to move first, followed by the rotation of the head. The eyes. moving with greater velocity, reach the signal first and then begin a backward movement with respect to the head to compensate for the continuing head rotation (Sanders, 1963; Bizzi, Kalil, and Tagliasco, 1971). The response is completed when t h e eyes a r e oriented directly toward the stimulus. The term “static fixation” refers to this cessation of both eye and head movement at the target (Robinson and Rath, 1977). It was thought previously that the purpose of this backward compensatory movement was to stabilize the system so that the image would remain on the fovea during the continuing rotation of the head. However, it was demonstrated for more complex visual stimuli (Bartz, 1966) that the backward compensatory movement occurred before the eye reached the target and must have been initiated by peripheral stimulation, rather than foveal stimulation. The targets used in this case were Arabic numerals, and the subjects were required to identify and vocalize the numerals. It seems reasonable to suppose that the process of target acquisition involves two levels of information processing during the course of the eye movement. One concerns the location of the target for initiating the backward compensatory movement, and the other invoIves the processing of information con-
HUMAN
FACTORS
tained in the target to enable an identification of the target itself. That is, when the eyes and head are moving towards a stimulus, enough information map be processed from the parafoveal retina to locate the target and enable an accurate backward compensatory movement. However, at this point in time, an identification of the target is not yet possible, and this processing must be delayed until the final foveal fixation (static fixation) on the target. On the basis of previous research it appears that little is known about the precise point in time when the processing of target content begins, and an identification of the target is possible (Senders, 1976). The present research was designed to provide information concerning this problem by determining the time course of the identification threshold during the backward compensatory movement. METHOD Apparat t IS
The apparatus consisted of the peripheral stimuli, subject earphones for masking extraneous noises, recording gear consisting of an amplifier, oscilloscope, and oscilloscope camera, and appropriate experimenter’s controls. The peripheral stimuli were arranged in a horizontal semicircle about the subject, and were at 60, 50, 40, and 20 degrees right and left of a fixation light. The subject was the center of a circle 1.83 m in diameter, and the eight stimuli and fixation light were at eye level. Each stimulus was a Burroughs type BDZOOS Nixie numerical indicator tube. The height of each numeral was 7.7 mm, subtending a visual angle of 14’ at the 1.83 rn distance. Of the ten digits, four (1,2,3, and 5) were chosen on the basis of preliminary tests as giving a good response for triggering the voice relay circuit. The tubes were mounted
June, 1979-333
ALBERT E. BARTZ
on a curved panel painted a flat gray to minimize glare. The experimental room was lighted by a fluorescent source placed 1.5 m above and 1.0 m behind the subject, and the luminance of the panel was 9.9 c a m ? . Eye movements were recorded by the electroculography method (EOG) of Mowrer, Ruch, and Miller (1936) and refined by Ford and Leonard (1 958). Beckman Biopotential Skin Electrodes (11 mm) were attached behind the external canthi of the subject's eyes, and the output of the electrodes was amplified and terminated at a dual-channel oscilloscope. The upper trace of the CRT was a record of the subject's eye mo\'ements, whiIe the lower trace indicated the onset and offset of the stimuli (see Figure 1). A camera was used to record the trace of the oscilloscope to critical stimuli. A decade timer was used to extinguish the numeral during various points in the subject's eye movement. The lowx trace of the dual-trace oscilloscope showed the input from this timer to enable a precise determination of the offset of the stimulus. Subjects
The paid subjects were three undergraduate men enrolled at Concordia College. They were examined by an ophthalmologist and were shown to be free from pertinent visual defects.
EYE MOVEMENT -I
STIMULUS
ON
L-I-
M -I-
P.F-I 1-P.0-
OFF
Figure 1. Typical rccordiiig shou*iiig eye riiorwzetit parameters and srbnzdzcs offset.
Procediire In both the training trials and the experimental trials, the procedure for the presentation of the stimuli was the same. The start of each trial was signaled by the experimenter by interrupting the 250 Hz, 60 db masking tone. At this signal the subject fixated the center fixation light. After an interval of Is, 2s, or 3s. a numeral in one of the indicator tubes came on, the fixation light was extinguished, and a timer (used only during the training trials) started. The subject then moved his eyes to the position of the stimulus and, during the training trials, verbalized the numeral into a microphone. This verbal response stopped the timer and extinguished the numeral in the indicator tube. During the experimental trials the stimulus was extinguished by the decade timer at various times during the subject's eye movement and the subject was then asked to guess the identity of the numeral just presented. As shown in'Figure 1, the upper trace (eye movement) remained horizontal until the subject began to move his eyes to the stimulus. As the subject started to move his eyes, the trace deviated from the horizontal axis. The eye movement reached its peak and began its backward (in the opposite direction) compensatory movement to compensate for the continuing head rotation. After the head had completed its rotation there was again a slight fonvard compensatory movement by the eyes before leveling off at the static fixation on the stimulus. The lower trace on the oscilloscope was a record of when the indicator numeral was on or off. As shown by Figure 1, the lower trace indicated that the numeral came on at the start of the trial and was extinguished sometime between the peak of the eye movement and its final fixation. Experiiimital Desigii
To ensure reliability of results, the subjects
334-June, 1979
were trained prior to the beginning of the experimental sessions by practicing responses to numerals at the 4 0 degree right and left positions. Subjects made 144 responses per day. 72 to each position. The training sessions were concluded on the 19th day when the means and the standard deviations became asymptotic. Four additional practice days were then held to familiarize the subjects with all eight stimulus positions. Also, the decade timer was used to estinguish the stimuli during the subject’s eye movements t o familiarize the subjects with this change in procedure. The experimental trials were run for 31 days and were initiated on the day following the last practice session. In order to study the information transmission during eye and head movement, the stimulus at 60 degrees left with the numeral 5 was selected as the critical stimulus. During the experimental trials the same stimulus presentation was used as in the training sessions, except that a total of 96 trials was presented instead of 144. There were three blocks of 32 trials each (8 stimuli x 4 numerals) with a rest period of 5 min between each block of trials. Each block contained all combinations of position and indicator numeral, and the combinations were randomized. Subjects performed as in the last four days of the training sessions, except that at irregular intervals the critical stimulus was inserted, and the subject’s rcsponse was recorded and photographed from the oscilloscope trace. Approximately 10-13 responses were recorded each day per subject. As in the last four days of the training trials, the.subject was alerted for each trial by an interruption of a tone in the earphones. One, two, or three seconds later, one of the numerals in one of the indicator tubes came on, and the subject, after moving his eyes and head to the stimulus, attempted to guess what numeral had been displayed. Sometimes the
HUMAN
FACTORS
numeral was extinguished so quickly that the subject had no idea what the numeral was. On other occasions the numeral remained on after the subject had completed his fixation on the stimulus. Regardless of how long the numeral was on, the subject was forced to guess the identity of the numeral. The subjects were unaware that only some of their responses were being recorded, i-e., the responses to the inserted critical stimulus. The photograph of the oscilloscope trace gave a precise indication of the time during the eye movement that the stimulus was extinguished. A typical record is shown in Figure 1, indicating the following parameters: L-latency, the time required for the eyes to begin moving. hl-movement time, the duration of the eye movement from the end of the latency to the peak of the eye movement. P to F-peak to fixation, the duration of the compensatory movement between the peak and the final static fixation. P to 0-peak to offset, the time between the peak of the eye movement and when the stimulus offset occurred.
For purposes of determining the ability of , the subject to recognize the stimulus, the P to 0 duration was the critical measure, since the subject’s response (correct o r incorrect) was tallied against the time of the offset of the stimulus. The EOG recording system had no appreciable lag time (
F A C T O R S , 1979,21(3),331-336
Information Transmission during Eye Movements ALBERT E. BARTZ', Departriiciit of Psycliolog): Coiicordia College, hfoorliead, Afiiiiiesota
Target stiiiiiili (the iiiiriieral 5 at 60 degrees to the left of the siibject's fixatioii point) i w - e extirigiiislied at varioiis tiiiies dirririg the siibject's eTe iiioiviiieiit to that stiiiiiiliis aiid the subject was asked to giiess what iiirrtieral had occiirred. Tlircshold frirrctioris were coiistnrcted from tliese data, arid it 1 1 ~foiriid s that at the begiririirig of the backward corripisatoty iiioveriierit, tary little iiifoniiatioii rt*astrarisriiitted. Hoit*e\vr,as tlie backi\wrd coiiipeiisatoty riioiwiieiit progressed, there litas ari iiicrease iri the correct ideiitificatioii of tlic stiriiirliis, arid a tliresliokd it*as reached before die fom*ardcoriipeiisatory iiiotwierit begati.
INTRODUCTION During the last decade there has been a renewed interest in the role of eye movements in visual perception due, in part, to the development of new techniques and the refinement of older methods for more precise eye movement recording. Research studies have ranged from descriptions of the physiology of eye movement control (e.g., Fuchs and Luschei, 1970; Skavenski and Robinson, 1973; Shebilske, 1976) to the role of eye movements in "higher" mental functions such as search behavior (Could, 1973; Mackworth, 1976). reading (Spragins, Lefton, and Fisher, 1976), and picture scanning and recognition (Noton and Stark, 1971; Antes, 1974). Basic to understanding the contribution of eye movements to a complex visual pcrceptual task is determining the manner in which information is processed during and after the eye movement itself. Several studies have investigated the change in the sensitivity of the Requests for reprints should be sent to Prof. Albert E. Bartz, Department of Psychology, Concordia College, hloorhead, hlinnesota 56560. U S A .
eye just prior to and during a saccadic movement (e.g.. Latour, 1962; Zuber and Stark, 1966; Volkmann, Schick, and Riggs, 1968; Ritter, 1976). In studies.of this type a short flash which is just above threshold (probability of detection = 1.0) for the stationary eye is presented at various times just prior to, during, and just after the voluntary eye movement. The subject then reports whether or not he saw the flash, and a large number of trials are run to determine the threshold from the detection function. Typical results usually show a saccadic siippressioii in that the detection threshold drops from 100% about 100 ms prior to the saccade to near 0% shortly after the saccade begins, and climbs gradually to 100% about 150 ms after the completion of the saccade (Volkmann, Schick, and Riggs, 1968; Volkmann, 1976).
Although the above studies have contributed greatly to the concept of visual sensitivity during eye movement, the detection of a threshold-level test flash represents a rather primitive level of functioning. In addition, the type of task needed to obtain the data re-
@ 1979, The Human Factors Society, Inc. All rights reserved.
33 1
332-June, 1979
quires the subject to make relatively restricted responses in that no head movements a r e permitted and the visual field to be searched is greatly reduced. When the subject is permitted to make a more “natural” response, i.e., he is permitted to move both head and eyes freely to fixate a peripheral stimulus that is more than 20 degrees from central fixation, the eye movement becomes more complex (Robinson, Koth, and Ringcnbach, 1976). It has been shown that at the onset of a randomly presented peripheral stimulus, the eyes begin to move first, followed by the rotation of the head. The eyes. moving with greater velocity, reach the signal first and then begin a backward movement with respect to the head to compensate for the continuing head rotation (Sanders, 1963; Bizzi, Kalil, and Tagliasco, 1971). The response is completed when t h e eyes a r e oriented directly toward the stimulus. The term “static fixation” refers to this cessation of both eye and head movement at the target (Robinson and Rath, 1977). It was thought previously that the purpose of this backward compensatory movement was to stabilize the system so that the image would remain on the fovea during the continuing rotation of the head. However, it was demonstrated for more complex visual stimuli (Bartz, 1966) that the backward compensatory movement occurred before the eye reached the target and must have been initiated by peripheral stimulation, rather than foveal stimulation. The targets used in this case were Arabic numerals, and the subjects were required to identify and vocalize the numerals. It seems reasonable to suppose that the process of target acquisition involves two levels of information processing during the course of the eye movement. One concerns the location of the target for initiating the backward compensatory movement, and the other invoIves the processing of information con-
HUMAN
FACTORS
tained in the target to enable an identification of the target itself. That is, when the eyes and head are moving towards a stimulus, enough information map be processed from the parafoveal retina to locate the target and enable an accurate backward compensatory movement. However, at this point in time, an identification of the target is not yet possible, and this processing must be delayed until the final foveal fixation (static fixation) on the target. On the basis of previous research it appears that little is known about the precise point in time when the processing of target content begins, and an identification of the target is possible (Senders, 1976). The present research was designed to provide information concerning this problem by determining the time course of the identification threshold during the backward compensatory movement. METHOD Apparat t IS
The apparatus consisted of the peripheral stimuli, subject earphones for masking extraneous noises, recording gear consisting of an amplifier, oscilloscope, and oscilloscope camera, and appropriate experimenter’s controls. The peripheral stimuli were arranged in a horizontal semicircle about the subject, and were at 60, 50, 40, and 20 degrees right and left of a fixation light. The subject was the center of a circle 1.83 m in diameter, and the eight stimuli and fixation light were at eye level. Each stimulus was a Burroughs type BDZOOS Nixie numerical indicator tube. The height of each numeral was 7.7 mm, subtending a visual angle of 14’ at the 1.83 rn distance. Of the ten digits, four (1,2,3, and 5) were chosen on the basis of preliminary tests as giving a good response for triggering the voice relay circuit. The tubes were mounted
June, 1979-333
ALBERT E. BARTZ
on a curved panel painted a flat gray to minimize glare. The experimental room was lighted by a fluorescent source placed 1.5 m above and 1.0 m behind the subject, and the luminance of the panel was 9.9 c a m ? . Eye movements were recorded by the electroculography method (EOG) of Mowrer, Ruch, and Miller (1936) and refined by Ford and Leonard (1 958). Beckman Biopotential Skin Electrodes (11 mm) were attached behind the external canthi of the subject's eyes, and the output of the electrodes was amplified and terminated at a dual-channel oscilloscope. The upper trace of the CRT was a record of the subject's eye mo\'ements, whiIe the lower trace indicated the onset and offset of the stimuli (see Figure 1). A camera was used to record the trace of the oscilloscope to critical stimuli. A decade timer was used to extinguish the numeral during various points in the subject's eye movement. The lowx trace of the dual-trace oscilloscope showed the input from this timer to enable a precise determination of the offset of the stimulus. Subjects
The paid subjects were three undergraduate men enrolled at Concordia College. They were examined by an ophthalmologist and were shown to be free from pertinent visual defects.
EYE MOVEMENT -I
STIMULUS
ON
L-I-
M -I-
P.F-I 1-P.0-
OFF
Figure 1. Typical rccordiiig shou*iiig eye riiorwzetit parameters and srbnzdzcs offset.
Procediire In both the training trials and the experimental trials, the procedure for the presentation of the stimuli was the same. The start of each trial was signaled by the experimenter by interrupting the 250 Hz, 60 db masking tone. At this signal the subject fixated the center fixation light. After an interval of Is, 2s, or 3s. a numeral in one of the indicator tubes came on, the fixation light was extinguished, and a timer (used only during the training trials) started. The subject then moved his eyes to the position of the stimulus and, during the training trials, verbalized the numeral into a microphone. This verbal response stopped the timer and extinguished the numeral in the indicator tube. During the experimental trials the stimulus was extinguished by the decade timer at various times during the subject's eye movement and the subject was then asked to guess the identity of the numeral just presented. As shown in'Figure 1, the upper trace (eye movement) remained horizontal until the subject began to move his eyes to the stimulus. As the subject started to move his eyes, the trace deviated from the horizontal axis. The eye movement reached its peak and began its backward (in the opposite direction) compensatory movement to compensate for the continuing head rotation. After the head had completed its rotation there was again a slight fonvard compensatory movement by the eyes before leveling off at the static fixation on the stimulus. The lower trace on the oscilloscope was a record of when the indicator numeral was on or off. As shown by Figure 1, the lower trace indicated that the numeral came on at the start of the trial and was extinguished sometime between the peak of the eye movement and its final fixation. Experiiimital Desigii
To ensure reliability of results, the subjects
334-June, 1979
were trained prior to the beginning of the experimental sessions by practicing responses to numerals at the 4 0 degree right and left positions. Subjects made 144 responses per day. 72 to each position. The training sessions were concluded on the 19th day when the means and the standard deviations became asymptotic. Four additional practice days were then held to familiarize the subjects with all eight stimulus positions. Also, the decade timer was used to estinguish the stimuli during the subject’s eye movements t o familiarize the subjects with this change in procedure. The experimental trials were run for 31 days and were initiated on the day following the last practice session. In order to study the information transmission during eye and head movement, the stimulus at 60 degrees left with the numeral 5 was selected as the critical stimulus. During the experimental trials the same stimulus presentation was used as in the training sessions, except that a total of 96 trials was presented instead of 144. There were three blocks of 32 trials each (8 stimuli x 4 numerals) with a rest period of 5 min between each block of trials. Each block contained all combinations of position and indicator numeral, and the combinations were randomized. Subjects performed as in the last four days of the training sessions, except that at irregular intervals the critical stimulus was inserted, and the subject’s rcsponse was recorded and photographed from the oscilloscope trace. Approximately 10-13 responses were recorded each day per subject. As in the last four days of the training trials, the.subject was alerted for each trial by an interruption of a tone in the earphones. One, two, or three seconds later, one of the numerals in one of the indicator tubes came on, and the subject, after moving his eyes and head to the stimulus, attempted to guess what numeral had been displayed. Sometimes the
HUMAN
FACTORS
numeral was extinguished so quickly that the subject had no idea what the numeral was. On other occasions the numeral remained on after the subject had completed his fixation on the stimulus. Regardless of how long the numeral was on, the subject was forced to guess the identity of the numeral. The subjects were unaware that only some of their responses were being recorded, i-e., the responses to the inserted critical stimulus. The photograph of the oscilloscope trace gave a precise indication of the time during the eye movement that the stimulus was extinguished. A typical record is shown in Figure 1, indicating the following parameters: L-latency, the time required for the eyes to begin moving. hl-movement time, the duration of the eye movement from the end of the latency to the peak of the eye movement. P to F-peak to fixation, the duration of the compensatory movement between the peak and the final static fixation. P to 0-peak to offset, the time between the peak of the eye movement and when the stimulus offset occurred.
For purposes of determining the ability of , the subject to recognize the stimulus, the P to 0 duration was the critical measure, since the subject’s response (correct o r incorrect) was tallied against the time of the offset of the stimulus. The EOG recording system had no appreciable lag time (
