Recording skin conductance (electrodermal response, galvanic skin resistance [GSR]) is a well-established technique for measuring autonomic arousal, which measures changes in electrical resistance of the eccrine sweat glands that are innervated by the sympathetic pathway. Electric shock is a powerful stressor even at harmless levels, and the arousal is readily detectable by GSR. We found that nystagmus beat amplitude and intensity increased during the contingent shock period (TD), but to a lesser extent during noncontingent shock (AA), and subsided in the final relaxed period (R;
Figs. 5a,
5c). We also found a similar decrease in the duration of FPs under stress. However, these measures are highly correlated and appear to reflect two underlying processes (factors 1 and 2). In factor 1, amplitude is the dominant factor associated with intensity and FPs, and is sensitive to the stressor. In factor 2, frequency is the dominant factor associated with intensity and FPs, but appears to be insensitive to the stressor. Further studies are needed to elaborate these factors. In any case, we concluded that INS is sensitive to sympathetic arousal, consistent with the findings of Cham et al.
19
It would seem intuitive that reduced FPs should reduce VA, but Cham et al.
19 reported no such reduction; however, their study was not designed to measure VA per se. In our study, we used standard optotypes, and found a small increase in VA during stress, which also was evident in the control group. We concluded, therefore, that the stress-induced increase in nystagmus intensity (and reduction in FPs) does not reduce VA. This result seems at odds with expectation and previous work. However, the majority of studies that have related VA or contrast sensitivity to specific nystagmus waveform parameters have not manipulated the nystagmus waveform within subjects, but only across subjects.
11,30–32 In biofeedback studies within subjects, VA has been reported to improve with reduced nystagmus, but the effect is small.
33 Even studies purported to predict VA from foveation
9,34 appear to be based on a between-subjects correlation, and not on changes in VA within individuals.
A possible explanation for the increase in VA with stress emerges when we examine RTs, which did increase significantly with stress (
Fig. 7). Consider, first, the control group. The median RT was approximately one second, implying that temporal summation extends far beyond the human visual integration time. It has been shown that partial temporal summation can occur up to 3 seconds or more.
35,36 Summation also can occur when stimulus exposure is split into two separate periods.
37 Adrian
36 modeled this extended speed-accuracy trade-off given by:
VA = 0.57 log
10 (
C ×
T) + 1.705, where
C is contrast and
T is exposure time (
C ×
T is total contrast dose). We propose that, when controls were punished for incorrect responses with contingent electric shock (TD), RTs increased because of a shift in criterion to more accurate responses, and, hence, increased acuity. If we assume that our response times include approximately 0.2 seconds movement time to press a button, then the mean exposure times (decision time) for the control group ranged from approximately 0.7 to approximately 1.0 seconds. Substituting into Adrian's formula, the increase in exposure time would yield an improved acuity of ∼0.025 LogMAR, for a high contrast Landolt C, which is in reasonable agreement with our observation (
Fig. 6).
For the INS group, the extended exposure can be accrued only across sequential FPs, and, hence, cycles of nystagmus. Since much time during a nystagmus cycle usually is spent outside the foveation window, overall RTs are much higher for nystagmats. As with the controls, there is a shift to longer exposure to improve accuracy with contingent shock leading to an increase in overall RTs. Cham et al.
19 also reported a similar magnitude of VA decrease in their restricted viewing task. However, the shock also acts as a stressor for the nystagmats, and reduces FPs (presumably via sympathetic pathways). Thus, even more nystagmus cycles are needed to accrue the requisite exposure time. Overall, this has some important ramifications.
First, if a nystagmat were to take unlimited time to respond, then VA should reach its maximum asymptotic value, which is determined by optics and neural pathways (refractive errors, sensory defects, amblyopia, and so forth). Manipulating the waveform (biofeedback, stress/relaxation, drugs, muscle surgery, null-eccentric viewing, and so forth) then will have only a negligible effect on VA, because the patient is at asymptotic VA. In reality, subjects (and controls) do not take infinite time, but will set a criterion depending on their understanding of task demand. However, if the subject expects to perform well without overt time restrictions (as occurs typically in a clinical setting), then VA will be only mildly affected by waveform manipulations, because the patient will increase or reduce their RTs to reach a desired accuracy. Under such manipulations (within subject), the durations of FPs would be only weakly or not correlated to VA, as observed empirically (see Introduction). In this scenario, VA may be a good measure of any underlying limit to the sensory pathway, but a relatively poor outcome measure in clinical trials that manipulate nystagmus waveform (e.g., drugs or muscle surgery).
On the other hand, if the patient is under pressure to respond quickly (e.g., driving, crossing roads, and so forth), or exposure time is restricted experimentally, then manipulating waveform may affect VA depending on the accrued foveation time. This is consistent with a study showing that restricted exposure time has a relative detrimental effect on VA in eccentric gaze for INS patients,
38 presumably because nystagmus is more intense (shorter FPs) when gaze is away from the average null zone. Increasing FPs would increase VA if the nystagmus cycle duration (beat frequency) remained unchanged, since there would be more foveal exposure time per second. However, if the cycle duration also increased, VA may be unaffected or even decrease. The problem is complex, depending on absolute FP durations, FP duty cycle, and underlying contrast sensitivity.
We emphasize that this study is not directly comparable to the “slow-to-see” phenomenon reported by Wang and Dell'Osso.
21,39 First, they recorded refixations to peripheral visual targets and measured the latency after a saccade of the first FP that was on target, and followed by successive accurate foveations. In our study, participants were fixating the central screen and were not required to make saccades (beyond their usual fast phases). Second, Wang and Dell'Osso
21 did not measure any perceptual component, but only examined eye movement traces. It is impossible to infer how long their subjects took to perceive or make a cognitive decision about the peripheral target. Based on our study, it can take a number of on-target FPs before a decision is made. We argue that their study reports “slowness to foveate a peripheral target,” and ours reports “slowness to make a visual decision.” The term “slow-to-see” is potentially misleading.
Our study has helped to clarify the role of stress in INS, but it raises an even more fundamental question. Waveforms can change dramatically in patients with INS, but this and previous studies, for example,
18,19 do not find a corresponding change in VA. As we have discussed, VA is largely insensitive to waveforms. However, the puzzle is why some studies report strong correlations between FPs and VA across subjects.
11,32,40 One possibility is that we have misunderstood the causal relationship between waveform and vision, such that VA is more or less fixed for an individual (due to sensory defects, amblyopia, unknown developmental reasons, and so forth), and that the waveform of a given subject adapts to the underlying VA. Wiggins et al.
18 reported some degree of plasticity, in which FPs increased with increased target resolution (“visual demand”). Thus, it is conceivable that the waveform is fine-tuned to optimize visual task performance. For example, as proposed previously, contrast for high spatial frequencies is optimized by long FPs, but not necessarily for low spatial frequencies.
18,41 For a VA task with unrestricted viewing, a nonoptimal waveform only has minor effects on VA, as VA is close to asymptote (see above), but this may not be the case for other tasks and clearly this issue must be explored further.
In summary, we propose that phrases, such as “effort to see,” “fixation effort,” “visual demand,” “task demand,” “slow-to-see,” are ill-defined and subject to misinterpretation. Even the term “stress” is difficult to define outside the context of physiologic sympathetic arousal. The use of these terms in the literature should, in future, be defined more carefully and precisely to test their implication properly for those with nystagmus. We concluded that sympathetic arousal (administered by electric shock and measured by SkC) strongly affects nystagmus waveforms and reduces FPs. However, this has little resulting impact on VA. The increase in RT seen with stress probably is a result of two mechanisms: (1) by shifting criteria to more accurate responses (and, hence, the slightly improved VA reported here), and (2) by shortening FPs and, hence, increasing the number of nystagmus cycles needed to reach the response criterion. This has implications for those with nystagmus in the real world as discussed above. We also suggested that VA alone is likely to be an insufficient measure of visual function for those with nystagmus. As a result, VA should be combined with other measures, such as “response time,” to give a more complete picture of visual performance. More research now is needed to investigate these findings further.