The Enright phenomenon is a binocular visual illusion of misperceived motion.1,2 The illusion can be induced by placing a neutral density filter over one eye of an observer looking sideways from a vehicle. When an observer is in this position one eye will be further forward in the vehicle and we define this as “the leading eye”. One eye will be closer to the rear of the vehicle and we define this as “the trailing eye”. If a filter is placed over the leading eye the observer's percept is that they are travelling faster than they actually are. If the filter is over the trailing eye the observer's percept is that they are travelling more slowly than they actually are. Enright, in his original paper2 on the illusion argued that it was related to the Pulfrich phenomenon,3 in which inter-ocular differences in retinal illumination cause misperceived depth of moving objects. The misperceived depth of objects moving across the visual field results in a disruption of motion constancy, which the observer misinterprets as travelling at a different speed.

The Enright Phenomenon is illustrated qualitatively in Figs. 1 and 2. Fig. 1 shows the Pulfrich Phenomenon,3 illustrating the case for an observer looking rightwards from a moving vehicle. The leading eye (the eye most forward in the vehicle) is the left eye in this situation. If the leading eye receives less retinal illumination than the trailing eye, as would be the situation where the trailing eye is dilated (as in Fig. 1a) the latency of perception will be slightly longer in the leading eye and objects in the field will appear slightly in front of those presented to the trailing eye, giving rise to uncrossed disparity, and observers will interpret these spurious depth cues as indicating that the object is further away. The reverse occurs when the leading eye receives more retinal illumination than the trailing eye, where the leading eye is dilated (as in Fig. 1b). The latency of perception will be slightly longer in the trailing eye and objects in the field will appear slightly in front of those presented to the leading eye, giving rise to crossed disparity, and participants will interpret these spurious depth cues as indicating that the object is closer. The effect of these spurious depth cues on velocity judgement is shown in Fig. 2, in which the observer is moving to the left causing objects to move across the visual field at a speed V. If the distance of objects is misjudged as being further away, then the apparent velocity V′ of objects moving across the field is interpreted as being larger than it actually is. If the observer judges their self-motion based on these visual cues then they will judge that they are travelling faster than they actually are. If apparent depth of objects is misjudged as being closer, then the apparent velocity V′ of objects moving across the field is interpreted as being smaller than it actually is. If the observer is judging their self-motion based on these visual cues they will perceive that they are travelling slower than they actually are.

Figure 1.

Schematic diagram of the Pulfrich phenomenon. Distortions of apparent depth for a moving observer induced by interocular asymmetry of retinal illuminance, with each eye perceiving the object in a slightly different relative spatial location, the disparity is interpreted as a spurious depth cue. The horizontal arrow indicates the movement of the observer (in this case to the left). (a) Trailing eye (in this case the right eye) receives more retinal illumination (shown unshaded): object is perceived as further away. (b) Leading eye (in this case the left eye) receives more retinal illumination: object is perceived as closer.

(0.08MB).

Figure 2.

The anomalous depth cues in the Pulfrich stereophenomenon can result in the actual velocity V being misinterpreted as an apparent velocity V′ which will be smaller if the perceived distance is smaller than it actually is (expected if more light enters the leading eye), and larger if the perceived distance is larger than it actually is (expected if more light enters the trailing eye).

(0.08MB).

The principles illustrated in Figs. 1 and 2 have been used to develop a geometric model of the Enright phenomenon which describes the relationship between interocular temporal latency differences and perceived self-velocity.1 These predictions have also been assessed quantitatively in a previous closed road driving study where participants were asked to drive at different speeds (60km/h or 40km/h), while looking sideways with 0.9 ND filters placed in front of the leading or trailing eye.1 When participants had a filter over the trailing eye they compensated for the percept of travelling slower by speeding up by an average of 8.7km/h, whereas when the filter was over the leading eye they compensated for the percept of travelling faster by slowing down by an average of 3.7km/h. This demonstrated that the Enright illusion can have real world consequences for driving behaviour. Carkeet et al.1 observed, however, that the real world speed distortions induced by the Enright Illusion were somewhat smaller than might be expected from their geometric model, hypothesising that cues such as vehicle vibration, engine noise, and monocular cues to depth, may all act to mitigate the real world consequences of the Enright phenomenon.

Apart from its consequences for driving, the Enright phenomenon has also been reported independently in the literature on electronic imaging,4 and has been considered as a possible reason for the crash of two Blackhawk helicopters.5

Although the traditional method for inducing the Enright phenomenon and the Pulfrich phenomenon has been to place a neutral filter over one of a participant's eyes, the Pulfrich Phenomena can also be induced by unilateral dilation of a participant's pupil6,7 or by restricting pupils through the use of apertures8 resulting in asymmetry of retinal illumination.9 Given its close relationship with the Pulfrich phenomenon it is likely that the Enright phenomenon can be induced by monocular pupil dilation, and may alter speed judgements while driving in this situation, however, this effect has not been previously explored.

Binocular pupil dilation has been shown to affect drivers’ ability to recognise and avoid low contrast hazards and also decreases visual acuity and contrast sensitivity.10 However, there have been no studies that have investigated the effects of monocular pupil dilation on driving performance. In addition, there appear to be no published studies on the frequency of monocular dilation for fundus examination. Nor could we find clinical guidelines or text books discussing whether monocular dilation is appropriate in ocular examinations. However, for centuries monocular dilation has been used, and is still used, as part of medical therapy for anterior uveitis and this use can continue for weeks to months in a patient.11,12 The Enright phenomenon is particularly relevant in those driving situations where drivers turn their head to one side for brief intervals when overtaking or judging safe gaps to change lanes or when navigating through intersections. In these situations monocular dilation could affect driving speed and speed judgments. Given the lack of information regarding how monocular pupil dilation affects driving, and the likelihood that monocular dilation could induce an Enright phenomenon, we conducted the current study to assess whether monocular pupil dilation could induce an Enright phenomenon and to determine the magnitude of such effects on real-world driving performance. We hypothesised that monocular dilation of the leading eye in the vehicle would cause the driver to perceive they are travelling slower than they actually are and that drivers would speed up to compensate and that monocular dilation of the trailing eye would cause drivers to perceive they were travelling faster than they actually are and slow down to compensate for the misperceived speed.

The group mean pupil diameters under the different testing conditions are shown in Table 1. Monocular pupil dilation induced an interocular difference in pupil area in log10 units averaging 1.02±0.14 SD with the right eye dilated and 0.99±0.12 SD with the left eye dilated. However, it is well recognised that, under photopic conditions, light entering close to the centre of the pupil is more effective at producing a visual response, than light entering the edge of the pupil. This is known as the Stiles-Crawford effect.17 Thus for the daylight conditions occurring in this experiment, the impact of the inter-ocular asymmetry of effective retinal illuminance may have been less than that predicted based on the relative ratios of pupil areas alone. Correcting pupil areas for the Stiles-Crawford effect, yielded estimates of interocular differences in effective retinal illuminance averaging 0.80±0.13log10 units with the right eye dilated and 0.75±0.11log10 units with the left eye dilated, using the pupil correction equation of Atchison and Smith:18

The group means for visual acuity under the different testing conditions are shown in Table 2. Visual acuity did not differ significantly for binocular viewing with an undilated pupil condition between day 1 and day 2 (t16=1.251, p=0.229), thus the visual acuity results reported and analysed are the mean of the two sessions. Visual acuity for the monocular viewing conditions did not differ between the groups that had their right eye or their left eye tested, (for the undilated condition t15=0.547, p=0.593; for the dilated condition t15=0.477, p=0.640). The visual acuity results analysed are the means of both groups. Repeated measures ANOVA showed a significant effect of viewing condition on visual acuity (F4,64=20.0, p<0.001) and post hoc analysis (matched pairs t-testing) showed that all pairs of viewing conditions in Table 2 had significantly different visual acuity, except for binocular dilated and monocular dilated conditions which did not differ significantly.

Main experimental results

Under binocular viewing conditions, monocular pupil dilation significantly affected driving speed when participants looked sideways from the vehicle. This is shown in Fig. 4 where there was a significant effect of Viewing Condition, (F3,48=8.974, p<0.001). In addition, there was a significant effect of Goal Speed (F1,16=308.447, p<0.001). There was no significant Viewing Condition x Goal Speed interaction on actual speed (F3,48=2.045, p=0.12). Post hoc testing showed that participants drove significantly faster when the leading eye was dilated compared to when both eyes were dilated (t16=2.57, p=0.02), but not significantly faster than when both eyes were undilated (t16=1.55, p=0.14). There was no significant difference in speed misjudgements when both eyes were either undilated or dilated (t16=1.058, p=0.322). Participants drove slower with the trailing eye dilated, than with both pupils undilated (t16=−4.11, p<0.001), with both pupils dilated (t16=−3.22, p<0.001), or with the leading eye pupil dilated (t16=−4.02, p<0.001).

Figure 4.

The actual speeds achieved under different binocular viewing conditions. Results are shown as group mean data (17 participants); Error bars are standard errors of the mean. Goal Speed of 40km/h is indicated by ●. Goal Speed of 60km/h is indicated by ○.

(0.07MB).

Control conditions

The control conditions (Fig. 5), assessing whether actual speed was differentially affected by binocular undilated viewing compared with monocular pupils dilated and undilated viewing, did not yield a significant effect of Viewing Condition (F2,30=0.593, p=0.559). Nor did the factor of monocular Viewing Eye affect actual speed, (F1,15=0.369, p=0.553) but there was a significant effect of Goal Speed on actual speed (F1,15=494, p<0.001), mirroring the results of the main experiment. There was no significant Viewing Condition×Viewing Eye Group×Goal Speed interaction (F2,30=0.706, p=0.502) nor were there significant interactions of Viewing condition×Goal Speed (F2,30=1.981, p=0.156), Viewing Condition×Viewing Eye Group (F2,30=1.606, p=0.217), Viewing Eye Group×Goal Speed (F1,15=3.949, p=0.065). Post hoc t-test analysis confirmed that there were no mean differences (p>0.05) between actual velocities for the right and left eye viewing groups irrespective of whether they were dilated or undilated, or had a Goal Speed of 60km/h or 40km/h.

Figure 5.

The actual speeds achieved under different control viewing conditions in the control experiments. Monocular viewing (dilated and undilated) is compared with binocular viewing undilated. Graphing conventions are as for Fig. 3. Results are means from 17 participants.

(0.07MB).

This study is the first to investigate the impact of monocular pupil dilation on driving perceptions and behaviour and demonstrates that monocular pupil dilation can induce distortions of driving speed, in accord with predictions based on the Enright phenomenon. This misjudgement of driving speeds occurs under specific driving conditions, with the observer's head turned sideways in the vehicle, looking perpendicular to the direction of motion of the vehicle, a situation which might arise during shoulder-checks on lane changes. The trend in the results confirmed those stated in the original hypotheses based on the Enright phenomenon. When the pupil is dilated in the trailing eye, retinal illuminance would be increased relative to the leading eye, which in the Enright phenomenon leads to the sensation of increased self-velocity. Hence drivers in the current study tended to slow their driving speeds, compared to the binocular pupil dilation and undilated conditions. Conversely, with the pupil dilated in the leading eye, the retinal illuminance would be less in the trailing eye. According to the Enright phenomenon, this would lead to a sensation of decreased self-velocity and hence it would be predicted that drivers would drive faster to compensate. The results we obtained are qualitatively similar to those of a previous study1 using similar methods, in which the Enright phenomenon was induced by using 0.9 ND filters to cause asymmetry of retinal illuminance. In the current study, the asymmetry in pupil size resulted in asymmetry of retinal illuminance equivalent to a 1.0 ND difference or, if corrected for the Stiles-Crawford effect, an equivalent difference of approximately 0.75 ND. The speed, with the leading eye dilated, was faster by an average of 7km/h than with the trailing eye dilated. In comparison, in the previous study using ND filters, the speed with the trailing eye filtered was faster by an average of 12km/h than with the leading eye filtered; a larger effect than the current study. There are a number of possible reasons for this difference in effect size. First, even with monocular pupil dilation, average retinal image illuminance will be substantially larger across both eyes than when neutral density filters are used. It is known that as average background adaptation increases, there is a decrease in the magnitude of luminance-induced -latency phenomena such as the Pulfrich phenomenon, Hess effect and simple reaction times.19 Like the Pulfrich phenomenon, the Enright phenomenon may depend on inter-ocular perceptual latency differences and may be similarly reduced by increasing background luminance.2 Second, it may be that differences in the retinal image quality arising from pupil dilation degrades the stereopsis cues used in the Enright phenomenon, although it should be noted that with clinical stereopsis tests, thresholds are not significantly affected by induced anisocoria or by 0.9 ND filters.20 Third, It is known from the previous study,1 that driving speed changes induced by the Enright phenomenon are much smaller than would be expected from the visual latencies likely to be induced by the inter-ocular asymmetries of retinal image brightness. Hence nonvisual cues (e.g. vibration and auditory cues), are likely to play a significant role in velocity judgements. Different vehicles were used in the current and previous studies and hence different nonvisual cues to velocity may have been available across the two studies, and contributed to the differences between the two studies in terms of the observed velocity changes.

Our control study, looking at the effects of monocular viewing on driving speed judgement, while looking sideways from the vehicle, gave largely negative results. Monocular viewing, either dilated or undilated, failed to result in a significant change in driving speed at either of the Goal Speeds, and there was no significant difference between the group that viewed with their left eye or the right eyes. We note however that because of limited access to testing facilities this was a mixed repeated measures/between groups design. A design where all factors were repeated across all participants would be a more powerful approach to investigate the effects of viewing with either the right eye or the left eye, and this may be a consideration for future research. Nonetheless, the control experiments support the theory that the anomalous speed judgements in the main experiment are dependent on binocular vision and are in agreement with Enright's explanation of the effect.2

Even with normal binocular viewing, participants tended to travel faster than their Goal Speed by an average of 5km/h for both the 40km/h Goal Speed and the 60km/h Goal Speed. This is in accord with previous research investigating speed estimation of drivers in normal traffic conditions, with speed underestimation errors of 4km/h at a speed of 40km/h and of 3km/h at a speed of 60km/h.5 On a closed circuit the participants in one study21 underestimated their speed by 7km/h at 60km/h. In comparison, a previous study1 of the Enright phenomenon, completed under similar conditions to the current study, showed that under conditions where retinal illumination was similar in each eye, participants matched their Goal Speed at 60km/h, but travelled faster than the Goal Speed of 40km/h by nearly 12km/h (7.5 miles/h).

Although monocular dilation is relatively uncommon, there are other potential sources of clinically induced anisocoria, for example unilateral corneal inlays such as Acufocus Kamra, which provide an artificially reduced pupil size to improve depth of field in presbyopia.22

It may be that such patients will experience perceived speed distortions found by the current study and clinicians should be aware of this as a potential complication of such procedures.

Visual acuity (with habitual driving spectacle correction) was affected by the different pupil dilation conditions. Worst acuity occurred under conditions in which the participant viewed through either one or both dilated pupils. Even under these conditions, acuity was still good, averaging 0.00logMAR (6/6). The worst acuity from a participant was 0.24logMAR (6/10.4), obtained under monocular viewing dilated conditions. This worst case visual acuity is still a line better than the 0.34logMAR (6/13.1) (Bailey-Lovie chart) which would be required to meet the Australian standards for a private driver's licence.23

When considering the implications of these findings it is important to note that the experiments were conducted under carefully controlled conditions on a closed track. The drivers achieved these speeds, looking sideways from a standing start, without access to information from the speedometer. The changes in velocity of 3–4km/h induced by the Enright effect were relatively small. Therefore it is worth considering how these findings potentially translate to driving in real world traffic conditions.

The changes in speed associated with monocular dilation of pupils occur under specific driving conditions, when the driver is looking sideways from a moving vehicle. Such conditions, though infrequent, do occur during normal driving, for example, when drivers are checking for other vehicles/road users before changing lanes. Such sideways head movements usually take place over a short interval of time, typically a number of seconds. If drivers are appropriately cautious in maintaining appropriate distances between themselves and traffic in front, then the velocity distortions induced by pupil dilation described here are unlikely to cause collisions. For example, if drivers are maintaining a 2s interval between cars, then the gap between vehicles would be 22m for traffic travelling at 40km/h. At such a distance if a car increased its speed by 4km/h, then it would close the gap to the car in front in 20s. For traffic travelling at 60km/h a 2s gap would correspond to a separation of 33m and if a car increased its speed by 4km/h the gap would be closed in 30s. However, there is evidence that, for motorway driving at these speeds, drivers leave a median spacing of approximately 1s between vehicles24 and so these times to contact would be reduced to 10s at 40km/h and 15s at 60km/h. Similar time frames to contact the car behind would be experienced if a driver slowed the car and the driver behind failed to take appropriate evasive action. It would be highly unusual for drivers to adopt such a sideways head posture for such a period of time. Thus the effects described in our research would be unlikely to be the sole factor contributing to an accident, but they are nonetheless important to be aware of and could be a contributing factor in more complicated accident scenarios.

In addition, the presence of the Enright phenomenon demonstrated in this research also suggests the potential for monocular dilation to induce other perceptual latency related stereophenomena which may affect driving. For example, the related stereoscopic anomaly, the Pulfrich phenomenon, might occur with a driver's head in a more normal posture (i.e. looking more or less straight ahead). Under these circumstances objects travelling past a driver's ear on the side with the undilated pupil would appear to swing in towards the driver as they pass, and objects passing on the side with the dilated pupil would tend to swing away.25 This may have implications for driving tasks such as lane keeping, so future research in this area is warranted in order to ensure that administering drops which result in monocular dilation do not impact on driving ability and safety.

In summary, this study shows that monocular dilation of pupils can affect driving speed perceptions and choices, in a manner that is consistent with previous studies of the Enright phenomenon. This effect is likely to have its origins in anomalous stereoscopic judgements of the depth of moving objects, arising under conditions where retinal illumination is different between eyes. Although previous researchers have reported that driving behaviour is affected by binocular pupil dilation, the current study is the first to report that monocular dilation of pupils can affect aspects of driving behaviour.

None declared.

None declared.