Short-term predictive changes in the dynamics of disparity vergence eye movements

Abstract

Repetitive stimulation of the disparity vergence system to large convergent step stimuli has been shown to increase the dynamics of subsequent responses to smaller step stimuli. Here we show that decreases in the dynamics of both disparity convergence and divergence eye movements can be induced using a frequently occurring small amplitude conditioning stimulus to modify responses to a larger, occasionally presented test stimulus. In one experiment, a simple conditioning stimulus consisting of repetitive 1° step stimuli was used to modify the dynamic vergence response to an occasional 4° step test stimulus. An experimental trial consisted of three phases: baseline, conditioning, and recovery. The baseline and recovery phases used only the 4° test stimuli. The dynamic characteristics of the responses to test stimuli were quantified by measuring the magnitude of the peak velocity. A statistically significant change was observed between the dynamics of conditioned responses compared to baseline and recovery responses indicting modification by the conditioning stimuli. During recovery, the response dynamics returned to levels near baseline levels showing that the decrease in response dynamics was caused by the conditioning stimulus, not fatigue. Another experiment showed that the response dynamics to large stimuli could be decreased whereas the dynamics of small stimuli could be increased by the same intermediate conditioning stimulus. Other experiments suggest that the modifications are due to a predictive mechanism. The results indicate that the dynamics of disparity vergence eye movements are malleable and depend to some extent on the amplitude of preceding stimuli.

Introduction

The vergence system is responsible for the convergence and divergence movements of the eyes. The first quantitative study of vergence dates to the work of Rashbass and Westheimer in 1961. Based on their experimental results, they modeled vergence as a simple feedback control system (Rashbass & Westheimer, 1961). However, more recent behavioral data support a dual control paradigm. Jones dissected the vergence response into a transient and sustaining component (Jones, 1980). The Dual-Mode Theory also describes vergence as a two-component system that includes a transient and a sustaining component (Hung, Semmlow, & Ciuffreda, 1986; Semmlow, Hung, & Ciuffreda, 1986) The transient portion is assumed to be an open-loop control component that enhances the speed of a vergence response. The sustaining portion is assumed to be driven by a visual feedback, closed-loop control system that provides fine tuning of the response to enable the extraordinary accuracy seen in binocular fixation. 

Investigators often use the terms prediction and adaptation in different ways. Prediction is a higher level process evoked when a stimulus is in some way predictable. In the life sciences, classical adaptation usually describes a process that requires a significant training period to change a behavior based on environmental conditions and that persists with a lasting aftereffect. From an engineering viewpoint, adaptive control is simply a modification of a system parameter irrespective of the time course of that change or its aftereffect. In some cases, adaptive control can be closely related to prediction (Deno, Crandall, Sherman, & Keller, 1995). 

Several investigations have studied oculomotor behavior in response to periodic stimuli where the timing, magnitude, and direction of the stimulus could be predicted. Prediction is typically marked by anticipatory movements as in the smooth pursuit system (Kowler, 1989), saccadic system (Kalesnykas & Hallet, 1987), and vergence (Alvarez, Semmlow, Yuan, & Munoz, 2002; Kumar, Han, Garbutt, & Leigh, 2002; Yuan, Semmlow, & Munoz, 2000b). In smooth pursuit, utilizing prediction led to a decrease in response latency and tracking error (Barnes & Asselman, 1991a; Barnes & Asselman, 1991b; Barnes, Barnes, & Chakraborti, 2000; Deno et al., 1995; Kowler, 1989). Finally, prediction will substantially reduce response latency in both saccadic and vergence eye movements (Krishnan, Farazian, & Stark, 1973; Rashbass & Westheimer, 1961). 

Adaptive behavior has also been observed in several oculomotor systems. The saccadic system can be manipulated utilizing a decreasing or an increasing gain protocol based on double step stimuli (Albano, 1996; Deubel, 1995; Semmlow, Gauthier, & Vercher, 1989; Straube & Deubel, 1995; Straube, Fuchs, Usher, & Robinson, 1997). The gain of the smooth pursuit system can also be modulated via adaptation (Churchland & Lisberger, 2002). The disparity vergence system also shows clear adaptive behavior: changes in the tonic vergence or phoria occurs in response to sustained demand (Patel & Firth, 2003; Schor, Gleason, Maxwell, & Lunn, 1993). Because this sustained demand is usually produced by prisms, this adaptive modification is termed “prism adaptation.” 

Recent studies have shown that disparity vergence dynamics can also be modified, but as we show here, these changes maybe due to a type of prediction (Munoz, Semmlow, Yuan, & Alvarez, 1999). Munoz et al. (1999) utilized an experimental protocol based on step-ramp stimuli (a step change in vergence immediately followed by a ramp) to increase the apparent gain of the system, resulting in convergence movements with large overshoots. Their findings indicate that the transient vergence component was strongly modified by these step-ramp stimuli. They also found a positive correlation between the maximum velocity of a subject's normal response and the increase in response velocity modified by the step-ramp conditioning stimuli (Munoz et al., 1999). Another study by Takagi et al. (2001) used a double-step protocol similar to that used in saccadic eye movements to both increase and decrease the gain. Studying only convergence eye movements, they showed that the peak velocity increased with an increasing gain protocol and peak velocity decreased with the decreasing gain protocol (Takagi et al., 2001). 

In the current study, four different experiments were performed: in three experiments, the stimulus direction, but not the timing, could be predicted; whereas in the fourth experiment, timing, direction, and amplitude could not be predicted. In all experiments, one or more “conditioning stimuli” were more prevalent than the stimulus used to “test” the modification. In the first three experiments, the dynamics of both convergence and divergence responses could be increased or decreased by conditioning using the appropriate repetitive step stimulus. One experiment showed that the dynamics of large responses could be decreased whereas the dynamics of small responses increased by the same intermediate conditioning stimulus. In the fourth experiment, when all predictive clues were eliminated, the absence of modification suggested that change in dynamics was the result of some type of prediction. Yet none of our stimuli were completely predictive because the stimulus onset was randomized. Moreover, the behavior observed was quite different from that seen in response to a truly predicable stimulus. Such stimuli, such as a periodic square wave, produce anticipatory eye movements and a decrease in latency, but not a change in movement dynamics. The modifications of this research show neither anticipatory movements nor latency changes but do show significant changes in movement dynamics. 

Methods

Subjects

Four subjects (20–58 years old) participated in this study. All subjects were male. All subjects signed informed consent forms before the experiments that were approved by the New Jersey Institute of Technology (NJIT) Institutional Review Board (IRB). During the experiment, the subjects' heads were immobilized using a custom chin rest to avoid any influence from the vestibular system. Subjects were instructed to initiate an experiment by depressing a button and to maintain binocular fixation on the stimulus. All subjects were able to perform the task with ease. One subject (Subject 001) was aware of the goals of this study and has been participating in eye movement experiments for many years. The other three subjects were naïve to the goals of the study and were inexperienced subjects. 

Visual stimulus

Disparity vergence stimuli were presented using a haploscope. Two analog oscilloscopes (BK Precision Model 2120B 30 MHz) were used to produce a symmetrical disparity vergence stimulus. Two partially reflective mirrors were placed in front of the subject's midline and it projected the two stereoscopically paired vertical lines from the oscilloscopes into the subject's line of sight. The stimulus displays were calibrated with real targets corresponding to 10° and 4° fixation points. Only the targets produced by the stimulus displays were seen by the subject, and there was no change in proximal cues associated with depth information as the target distance was constant. 

For all experiments, accommodation was held constant. The stimuli were not seen through pinholes, thus accommodation was not open loop. The experiments did use a haploscope; consequently, stimuli appeared at the same optical distance from the subject. During all experiments, the initial vergence angle for the stimuli was 8° for all four subjects. The stimulus displays were placed 56 cm away from the subject. The targets were 3 cm × 2 mm, which remained constant throughout the experiment. The stimuli were symmetrically presented on two oscilloscopes where the subject's head was positioned so that the targets appeared along the midline. 

The experiment was designed to generate two types of stimuli—a test stimulus and a conditioning stimulus. This study investigated how the conditioning stimulus influenced the dynamic properties of the response to the test stimulus. 

Experimental protocol

The experimental protocol was based on that developed by Munoz et al. (1999). Two experimental modalities—predictive and nonpredictive—were utilized. During the predictive experiments, convergence and divergence stimuli were presented in separate experimental sessions utilizing a noninterleaved format. In this session, the subject could predict the direction (either inward or outward), but not the timing, of the stimulus. During the nonpredictive experiments, the convergence and divergence stimuli were interleaved; thus, the subject could not anticipate the direction or timing of the stimulus. In all four experiments, the stimulus was presented following a random delay between 0.5 and 2.0 s from the time the subject indicated readiness by depressing a button. This delay was sufficient to eliminate subject prediction to timing information. After the random delay, the step stimulus was presented for 3 s. Eye movement responses were recorded during the 3-s stimulus presentations as described below. 

An experimental run consisted of three phases: baseline, conditioning, and recovery. The baseline phase consisted of only test steps and served as the control. The next phase, conditioning, consisted of the conditioning stimulus with an occasional, randomly occurring test stimulus. The amplitude of the conditioning and test stimuli varied with the particular experiment as described below. The stimulus presentation was also randomized so that on average the subject was five times more likely to view the conditioning stimulus. In the first three experiments, the subject could predict the stimulus direction (inward or outward) and knew that the majority of the time the stimulus would be the conditioning step stimulus. Munoz et al. (1999) described the time course of the modification and showed that changes occur very rapidly once the conditioning stimulus was applied. Accordingly, we average the dynamic response of the test stimulus beginning with the conditioning phase. The conditioning phase generally lasted long enough to collect 10 test responses and was limited by the desire to keep fatigue to a minimum (Yuan & Semmlow, 2000a). The last phase of the experiment was the recovery phase, which was used to determine if fatigue was influencing the results. Munoz et al. (1999) have shown that the modification has very little aftereffect and quickly returns to some steady level. As with the baseline phase, the recovery phase contained only test stimuli. 

Approximately 10 test stimuli were collected during each phase during a typical experimental session. During conditioning, at least 50 conditioning stimuli were required to attain the 10 test stimuli. Each experimental session lasted approximately 30–45 min. Usually three or four sessions were needed to obtain a sufficient number of artifact-free test responses. 

In the first experiment, the conditioning step of 1° was smaller than the test step of 4°. The second experiment investigated if the same conditioning stimulus (1° step) would produce a dynamic change in larger test responses (8° step). In the third experiment, test steps of 1° and 4° were conditioned by an intermediate 2° conditioning stimulus to determine if both increases and decreases in movement dynamics could be achieved using a single conditioning stimulus. Given the stimulus and recording challenges of these two experiments (responding to 8° stimuli without excessive saccades and recording small 1° responses), only the experienced subject (001) could effectively participate. The fourth experiment presented convergence and divergence stimuli in an interleaved format so that the subject could not predict the stimulus in terms of timing, direction, or magnitude, although the subject still knew that the conditioning stimuli were more likely to occur. Test steps of 1° and 5°, both convergent and divergent, were conditioned by 3° convergent and divergent conditioning stimuli. The various experimental conditions for the four experiments are summarized in Table 1. 

Table 1

 

Overview of the four types of experiments: Experiments 1 through 3 presented convergent and divergent stimuli in separate sessions, thus the subject could predict direction. In Experiment 4, the convergent and divergent stimuli were both presented during the same experimental session, thus the subject could not predict the direction of the stimulus. Convergent stimuli are noted as positive, whereas divergent stimuli are denoted as negative.

Experiment number	Predictive ability	Conditioning stimulus	Test stimulus
1	Direction and size	+1	+4
		−1	−4
2	Direction and size	+1	+8
3	Direction	+2	+1, +4
4	None	+3, −3	+1, +5, −1, −5

Eye movement recording

Eye movements were recorded using an infrared limbus tracking system (λ = 950 nm) manufactured by Skalar Iris (model 6500). The manufacturer reports the system has a resolution of 2 min of arc and a bandwidth of 100 Hz. All eye movements were well within the system's reported ±25° linear range. The left and right eye movements were recorded and saved separately. The presentation of stimuli and the digitization of signals that were saved to disk were controlled by a custom LabVIEW program. Data acquisition was done at a sampling rate of 200 Hz, which is well above the Nyquist frequency for vergence eye movements. Calibration of left and right eye movement responses was performed by recording the output of the eye movement monitor at two known positions before and after each response. Calibration data for each eye were stored with the response and used to construct the eye movement response during offline data analysis. 

Data analysis

Data analysis began by converting raw digitized left and right responses to degrees using the calibration data. The left and right eye movements were inspected individually and responses that contained blinks or saccades in the transient portion of the response were deleted. The main objective of this study was to investigate the dynamic change in the responses. Therefore, responses that contained small saccades during the final, steady-state portion of the movement were analyzed. The left and right eye responses were subtracted to yield the net disparity vergence response where convergence is plotted as positive and divergence is plotted as negative. The velocity response was computed using a two-point central difference algorithm (Bahill, Kallman, & Lieberman, 1982). 

Data were analyzed by measuring the magnitude of the peak velocity and compared using an unpaired student t test to determine if the dynamic changes were statistically significant. Data were analyzed using MATLAB (Waltham, MA) and were plotted and statistically analyzed using the software package Axum (Cambridge, MA). For all figures, the averaged response is plotted where the responses were aligned based on peak velocity and then averaged. 

Results

Averaged convergence eye movements are shown for two subjects in Figure 1. All responses are identical to 4° step changes in vergence stimulus. The upper traces are velocity (degrees per second) and the lower traces are vergence position (degrees). The blue line shows the averaged responses from the baseline phase that involved only 4° step stimuli. The red lines are also 4° averaged responses but were recorded during the conditioning phase when the subject was presented with five times as many 1° responses as 4° responses. The green lines are 4° averaged responses recorded during the recovery phase. 

Figure 1

 

Experiment type 1 where the subject could predict direction and the subject knew he was more likely to view the 1° convergence step during the conditioning phase. The upper plots are averaged convergence responses from baseline, conditioning, and recovery phases of the experiment for two subjects. In the upper plots, the bottom traces are vergence position (difference in the left and right eye movement where convergence is positive) and the top traces are velocity. The baseline, conditioned, and recovery averaged responses are denoted as blue, red, and green lines, respectively. The number of samples in the average plots corresponds to the numbers listed in Table 2. The lower plots show the peak velocity as a function of trial number for each of the 4° test responses from the baseline, conditioned, and recovery phases where the circles, triangles, squares, and diamonds are responses recorded from Days 1 through 4, respectively. The solid lines represent the average peak velocity from the baseline, modification, and recovery phases denoted as blue, red, and green, respectively.

Test responses collected during the conditioning phase show slower transients with decreased peak velocities and an increase in the duration of the response (note particularly the velocity traces in Figure 1). All responses eventually attain the final stimulus position of 4°, but conditioned responses require more time to reach the final position. The magnitude of the peak velocity for convergence movements decreased from the mean peak velocity of the baseline responses by up to 40% during conditioning (see Table 2). 

Table 2

 

Peak velocity mean (degrees per second), SD, and number of samples for Experiment type 1 where the subject could predict stimulus direction and the subject knew he was more likely to view the 1° step during the conditioning phase.

Subject	Experimental protocol	Magnitude peak velocity mean (degrees per second) ± SD	Number of samples
Convergence
001	Baseline	27.1 ± 3.66	27
	Conditioning	23.3 ± 4.44	20
	Recovery	30.2 ± 5.92	31
002	Baseline	16.4 ± 4.39	23
	Conditioning	9.81 ± 2.62	17
	Recovery	12.9 ± 2.18	23
003	Baseline	18.1 ± 2.62	23
	Conditioning	14.5 ± 2.05	13
	Recovery	16.3 ± 1.69	22
004	Baseline	16.3 ± 3.95	30
	Conditioning	13.7 ± 2.11	16
	Recovery	16.0 ± 3.85	15
Divergence
001	Baseline	12.6 ± 2.21	43
	Conditioning	10.8 ± 2.60	27
	Recovery	14.1 ± 4.24	36
002	Baseline	13.2 ± 3.74	39
	Conditioning	10.6 ± 4.15	17
	Recovery	12.9 ± 3.39	29
003	Baseline	11.3 ± 3.25	27
	Conditioning	9.08 ± 1.58	18
	Recovery	12.0 ± 4.50	28
004	Baseline	16.9 ± 5.59	33
	Conditioning	12.8 ± 3.30	27
	Recovery	14.6 ± 3.29	29

Averaged divergence data are plotted in Figure 2 in the same format used for convergent responses in Figure 1. Divergence responses observed during the conditioning phase also showed slower dynamics as seen by a decrease in the magnitude of peak velocity and an increase in the duration of the response. The magnitude of the peak velocity for divergence decreased during conditioning by up to 24% (see Table 2). 

Figure 2

 

Experiment type 1 where the subject could predict direction and the subject knew he was more likely to view the 1° divergence step during the conditioning phase. The upper plots are averaged divergence responses from baseline, conditioning, and recovery phases of the experiment for two subjects. In the upper graphs, the bottom traces are position (difference in the left and right eye movement where divergence is negative) and the top traces are velocity. The baseline, conditioned, and recovery averaged responses are denoted as blue, red, and green lines, respectively. The number of samples in the average plots corresponds to the numbers listed in Table 2. The lower plots show the peak velocity as a function of trial number for each of the 4° responses from the baseline, conditioned, and recovery phases where the circles, triangles, squares, and diamonds are responses recorded from Days 1 through 4, respectively. The solid lines represent the average peak velocity from the baseline, modification, and recovery phases denoted as blue, red, and green, respectively.

For all subjects, the magnitude of peak velocity decreased during the conditioning phase compared to the baseline and recovery responses for both convergence and divergence eye movements ( Table 2). The magnitude of peak velocity measured from the recovery responses was approximately equal, or occasionally less than, the peak velocity from the baseline responses. Nonetheless, the average peak velocity from responses in recovery was always greater than that measured from conditioning, indicating that the dynamics were altered due to the stimulus conditions, not fatigue. A statistical analysis revealed that the reduction in peak velocity in the conditioning phase, compared to the baseline or recovery phase, was significant ( p < .05), Table 3. 

Table 3

 

Statistical comparison of peak velocity for results presented in Table 2.

Subject	P value base vs. conditioning	P value conditioning vs. recovery
Convergence
001	.0024	<.0001
002	<.0001	.0002
003	.0002	.0082
004	.0186	.0464
Divergence
001	.0028	.0007
002	.0245	.0470
003	.0102	.0114
004	.0014	.0460

The lower plots in Figure 1 plot each test trial's peak velocity as a function of the trial number recorded on a given day. Two subjects' time plots are shown in Figure 1 for convergence and Figure 2 for divergence. The data collected from three to four experimental sessions are overlaid where each day is denoted by a different symbol. We did not observe any trends in the data comparing 1 day to another. The response dynamics are quickly modified during the conditioning phase compared to baseline and similarly the dynamics return to baseline levels quickly during the recovery phase, see Figures 1 and 2. 

Modification of the vergence response is also found in one subject (001) using a larger test stimulus of 8°. In this experiment, an 8° test step was modified by the 1° conditioning stimulus. Results show that the changes between baseline and conditioning were statistically significant ( p < .02), Table 4. This experiment indicates that this behavioral change occurs for different vergence step amplitudes. 

Table 4

 

Results of Experiments 2 and 3 where the subject could predict direction of the stimulus and knew he was more likely to view the conditioning stimulus during the conditioning phase of the experiment.

Experiment		Peak velocity (degrees per second)
Conditioning step stimulus	Test step stimulus	Baseline response	Conditioned response	Recovery response
1°	8°	50.0 ± 6.9 ( n = 33)	44.7 ±7.0 ( n = 15)	53.2 ± 6.3 ( n = 16)
2°	1°	8.0 ± 2.2 ( n = 10)	10.3 ± 1.7 ( n = 12)	8.6 ± 2.3 ( n = 8)
	4°	25.1 ± 4.9 ( n = 23)	20.1 ± 4.1 ( n = 15)	24.7 ± 3.7 ( n = 11)

In the third experiment, the vergence system was observed when a subject was presented with two different test stimuli, one of larger magnitude and another of smaller magnitude than the conditioning stimulus. Averaged behavioral data of the test responses (4° and 1°) can be seen in Figure 3 where the averaged baseline responses are plotted using a blue line, the averaged conditioned data are plotted using a red line. In the presence of a 2° conditioning stimulus, the 1° test response showed an increase in dynamics whereas the 4° test response showed a decrease in dynamics indicated by peak velocity, Table 4. The comparison between baseline and conditioned peak velocities showed statistically significant results for both the 1° test responses ( p < .015) and 4° test responses ( p = .002). These results show that a conditioning stimulus can simultaneously alter the vergence dynamics of two different convergence test stimuli: one that is larger and another that is smaller than the conditioning stimulus. 

Figure 3

 

Experiment type 3 where the subject could predict direction. Averaged 4° and 1° step responses where position is plotted in the lower traces and velocity is plotted in the upper traces. The baseline, conditioned, and recovery averaged responses are denoted as blue, red, and green lines, respectively. Data show that a conditioning stimulus (2°) can simultaneously decrease the dynamics of a larger test response (4°, left plot) while increasing the dynamics of a smaller test response (1°, right plot). The left and right graphs have different scales. The number of samples in the average plots corresponds to the numbers listed in Table 4.

The fourth experiment, which contained four test stimuli presented within the same experimental session (1°and 5° converging and diverging steps), did not exhibit a significant change in dynamics when comparing the test responses from the conditioning phases to those recorded during the baseline phase for either subject studied. 

Discussion

The results demonstrate that the stimulus environment can influence the responses of both convergence and divergence movements. This was true for experienced and inexperienced subjects. During the conditioning phase, the response dynamics decreased up to 40% indicating that the conditioning stimulus, which was one-fourth the magnitude of the test response (4° step), modified the vergence eye movement response. The rapid changes do not follow the typical temporal pattern seen in other forms of adaptation. The modification occurs quickly and the aftereffect is also brief as the response quickly returns to the baseline dynamics, see Figures 1 and 2. 

The experiment with two test stimuli (one of greater magnitude than the conditioning stimulus and one of lesser magnitude) reduced the subject's ability to predict the amplitude of the stimulus. However, the stimulus direction was always the same, either all converging or all diverging. In addition, the conditioning stimulus was still the most likely stimulus to be presented. This prevalence resulted in a modification of both the larger and smaller test stimuli. 

The last experiment performed reduced prediction further by randomizing the stimulus onset, direction, and magnitude of the test stimulus, although the conditioning stimulus amplitude was still more prevalent. Two subjects participated in this experiment. Results show that during the conditioning phase, on average, the test responses did not exhibit a substantial change in dynamics compared to the baseline responses. During the last experiment, when the four test stimuli and the two conditioning stimuli were presented, the modification of dynamics was not observed. In this experiment, the direction, magnitude, and timing information was not predictable by the subject. This suggests that some type of predictive mechanism produces the changes in response dynamics observed in the other three types of experiments. However, none of the stimuli were completely predictable because neither their onset times nor amplitude was known, although the conditioning stimulus amplitude was more likely. Moreover, none of the responses showed the decrease in latency or the anticipatory premovement drifts found in vergence responses to truly predictable stimuli (Alvarez et al., 2002; Yuan et al., 2000b). Hence, findings suggest a predictive process that is somewhat different from that seen in response to stimuli that are predictable in both amplitude and time. 

The dynamic modifications observed in this study occurred rapidly and quickly returned to baseline conditions during the recovery phase; see the time course plots in Figures 1 and 2. For most subjects, the dynamics observed in the recovery phase were somewhat less than the baseline dynamics, which we attribute to fatigue. Other researchers have noted a decrease in peak velocity of 20% after 100 repetitive eye movements resulting from fatigue (Yuan & Semmlow, 2000a). Nonetheless, the recovery phase revealed some increase in dynamics over conditioned responses even for those subjects who showed some fatigue. Convergence responses are generally faster than divergence responses and this was the case for three of our four subjects. The modification was also larger in convergence than in divergence, Table 2. 

Future studies will involve new signal processing algorithms such as Independent Component Analysis to decompose the combined vergence signal into its transient and sustained components to further investigate the relationship between the two (Semmlow & Yuan, 2002a; Semmlow & Yuan, 2002b). 

Conclusion

The dynamics of convergence and divergence disparity responses can be reduced when the subject is more likely to view a stimulus of lesser magnitude. The modifications were quantified by peak velocities and showed statistically significant changes in conditioned responses. Convergence movements exhibited a greater change than divergence movements, which may be related to the difference in the magnitude of the transient components in the two systems. 

Acknowledgments

This work was supported in part by funds from Essilor International and NSF CAREER BES 0447713. 

Commercial relationships: none. 

Corresponding author: Tara L. Alvarez. 

Email: [e-mail]. 

Address: Department of Biomedical Engineering, New Jersey Institute of Technology, University Heights, Newark, NJ 07102.