U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590
202-366-4000


Skip to content
Facebook iconYouTube iconTwitter iconFlickr iconLinkedInInstagram

Federal Highway Administration Research and Technology
Coordinating, Developing, and Delivering Highway Transportation Innovations

 
REPORT
This report is an archived publication and may contain dated technical, contact, and link information
Back to Publication List        
Publication Number:  FHWA-HRT-15-027    Date:  November 2015
Publication Number: FHWA-HRT-15-027
Date: November 2015

 

Information As A Source of Distraction

 

Chapter 1. Introduction to the Information as a Source of Distraction Project

The overall goal of the Information as a Source of Distraction project was to further the scientific basis for decisions about the types of information that can be displayed within the right-of-way without adversely affecting drivers’ attention to their primary task—safe driving. The goal was to be accomplished by evaluating the distraction potential of various types of information when displayed on electronic changeable message highway signs (CMSs), especially non-traffic-related messages displayed within the right-of-way. In addition to CMS messaging, the distraction potential of frequent and closely spaced freeway navigation guide signs was to be considered.

Objectives

The objectives of this project were to determine the distraction potential of various types of information when displayed within the right-of-way. In particular, the studies were to determine the potential of CMS depictions of road signs and non-traffic-related messages and other information sources and to determine the distraction potential of guide signs that are more closely spaced or more frequent than current criteria permit. The objectives were to be achieved through performance of both on-road and driving simulator evaluations. The researchers were to create a report and present recommendations with a scientific basis to enable practitioners to assess the informational load imposed on road users by information sources within the highway right-of-way.

Background

CMSs are used by highway agencies to communicate current information to drivers.[1] Information displayed on these signs may include travel times, incident information, temporary lane restrictions, and alternate route suggestions. CMSs are intended to aid motorists, transportation agencies, and the general public.(2) The signs are thought to benefit motorists by increasing safety and reducing travel delays. Transportation agencies are thought to benefit from improvements in network operational efficiencies. The general public is thought to benefit from reductions in pollution that result from network efficiencies.

Because CMS display technology continues to evolve and improve, the potential for the development of novel CMS uses is likely to continue. It is also conceivable that some of the extensive guidance for CMS operations and policy may become obsolete as technology continues to evolve.(3–5) The FHWA has requested additional research to assist future decision making, and guide validation or update of CMS standards and guidelines contained in the Manual on Uniform Traffic Control Devices for Streets and Highways (MUTCD).(2) The present project is intended to serve that role, with an emphasis on guidance to avoid unnecessary driver distraction by CMS content. This literature review is intended to summarize the current knowledge base regarding CMS information display and its relation to driver distraction.

Distraction is a construct that is related to attention and stimuli that inappropriately divert or attract attention. Before distraction can be studied in a scientific manner, an operational definition is required.

Operational Definition of Distraction

In their book on driver distraction, Regan, Lee, and Young define distraction as “…a diversion of attention away from activities critical for safe driving toward a competing activity.”(6) They go on to say the following:

Because driving involves the management of multiple tasks, such as monitoring the road ahead, monitoring the speedometer, monitoring the review mirrors, and perhaps reading safety-critical messages on a CMS, Regan et al. qualify distractions as an “inappropriate” distribution of attention.(6)

The US-EU Bilateral ITS Technical Force working group on driver distraction provided a similar definition of driver distraction to that provided by Regan, Lee, and Young.(7) The members of the working group also distinguished between appropriate and inappropriate attention. They acknowledge a rear-end collision with a suddenly braking vehicle would not be caused by distraction if the driver of the following vehicle was performing a shoulder check while merging with mainline freeway traffic. That is, when two safety-critical demands for attention are simultaneous, workload—not distraction—would be the main causative factor for a collision. The working group defined critical for safe driving tasks as those that allow the driver to avoid or not cause a crash. The working group acknowledged that safety-critical tasks may be situation dependent and liable to interpretation. They caution against using hindsight (i.e., whether a crash actually occurred) as a criterion for determining safe driving. As yet, however, there are no universally accepted metrics for assessing safe driving in the absence of a crash.

Horberry and Edquist discuss distraction from elements outside of the vehicle.(8) Their analysis focuses on billboards and the potential for billboards to attract driver attention from the built roadway, which they define as roadway geometry, the roadway surface, and traffic signs and markings. Despite this narrow scope, they conclude that although it seems intuitively obvious that billboards are a distraction to drivers, there is little research to support or refute this view. Their analysis provides little guidance on an approach to the distraction potential for CMSs that are part of the built roadway. However, their analysis does provide suggestions for the direction of future research. They identify the following three types of distracting stimuli:(8)

Visual distraction occurs when a strong visual stimulus causes an involuntary glance, as might occur in response to a bright flash of light. Cognitive distraction occurs when a stimulus causes the driver to inappropriately devote mental attention to a matter not immediately necessary to safe driving, such as engaging in a conversation. Secondary activities are non-driving-related activities in which drivers voluntarily engage, such as searching for a vehicle identified in an AMBER alert that was posted on a CMS.

Thus, in the study of the relationship between CMS content and driver distraction, it is suggested that researchers are looking at the content of CMS that results in inappropriate diversion of attention away from activity critical for safe driving to CMS stimuli that do any of the following:

There are several challenges to successfully addressing the information distraction issue. CMSs on freeways are primarily, if not exclusively, traffic control devices (TCDs). Thus by definition, CMSs are part of the roadway, and reading them is part of the driving task. In some cases, CMSs contain information critical to safe driving. This implies that a CMS, when used as a TCD, can only be distracting if it causes an inappropriate diversion of attention (i.e., when the information on the CMS is not more critical to safe driving than other roadway visual stimuli). When used to display non-traffic-related messages, e.g., advertising, AMBER alerts, or public service announcements (e.g., click it or ticket), the bar for distraction is lower (i.e., this information is not safety critical) and any attention that would otherwise be devoted to safety critical tasks would then constitute distraction. Even in the case of non-traffic-related messages however, there is room for diversion of driver attention, if that diversion does not detract from safety-critical activity.

The amount of driver mental capacity available for safety-critical driving tasks can vary from time to time, person to person, and place to place. For instance, required capacity can vary in time with the level of traffic. Persons who are fatigued may have less available capacity.(8) Freeway sections with sharp curves, narrow lanes, or frequent weaving generally require more attention than straight, wide, uninterrupted sections. Therefore, assessing the distraction effects of CMS information requires either methods of measuring attention requirements and available driver attention capacity or the use of some putative surrogate measure of distraction.

Measuring Attention and Distraction

Young, Regan, and Lee discuss various driving performance measures to assess the distraction effects.(9) They mention lateral (lane) tracking performance, longitudinal (following headway) performance, gap acceptance, steering entropy, reaction time, and speed maintenance as potential driving performance measures that have been shown to vary with putative levels of distraction. They caution that these measures are not interchangeable.

Visual distractions (tasks that require looking away from the roadway) have been shown to affect lateral tracking performance, whereas tasks with moderate levels of cognitive distraction are associated with improved lateral control (measured by lane tracking and steering entropy). Reaction times to traffic events have been shown to be degraded by cognitive distraction. Young, Regan, and Lee also reported that in many cases driving simulations are more sensitive to driving performance degradation than are on road and test track assessments.(9) However, they caution that this greater sensitivity in driving simulations may be the result of shifts in drivers’ criterion for attending to the driving and distraction tasks (i.e., in simulators drivers may give higher priority to attending to the distraction and lower priority in attending to the driving task). How this shift in priorities relates to driving safety in the real world is unknown. However, this phenomenon points to the importance of measuring both primary (driving) and secondary (cognitive) task performance to assess how drivers are trading off performance in the two tasks.

Where drivers look can also be a clue to where their attention is directed.(10) Although it is possible to attend to locations on which the eyes are not focused and the level of attention to the location where the gaze is fixed can vary because of cognitive distraction, the center of gaze is generally a good indication of where drivers are attending.(11,12) Furthermore, there is a long history of using eye-tracking technology in the study of how drivers attend to traffic signs.(13,14)

The fovea is the area of the retina that has the highest resolution and is used for reading normal text. The fovea covers about 2 degrees of visual angle. To capture fine detail, the direction of gaze must change to place the detail within the fovea. Assuming good contrast, an individual with 20/20 visual acuity can read a highway sign with letters 18 inches (46 cm) high from 1,000ft (305 m) away. At 1,000 ft (305 m), a 2-degree cone of vision would encompass a diameter of 36 ft (11m). A CMS of 8 ft (2.4 m) in height (typical of a gantry-mounted overhead CMS on freeways) with the bottom of the sign at 20 ft (6.1 m) above the roadway would fall within this 2‑degree cone, even if the driver is focused on a vehicle on the roadway below the sign. Furthermore, when deployed in a moving vehicle, well-calibrated eye trackers are limited to about 1.5 degrees of resolution.(15) As a result, a field study with an eye tracker may not be able to establish with certainty when a driver first begins reading an overhead freeway sign or when the first glance at the sign occurs. At 500ft (152 m), the 2-degree cone of fine vision has an 18-ft (5.2-m) diameter, so at that distance, it should be possible to determine with reasonable certainty whether a driver’s eyes are directed toward an overhead sign or to the roadway ahead. However, at 500 ft (152 m), another complication arises. The 18-inch (46-cm) letter will subtend 10 min of arc or about twice the angle required for legibility. Text of this size, especially if it is expected, such as a route number or speed limit, would be legible in near peripheral vision (2 to 10 degrees from the center gaze). This means that a driver may comprehend a CMS message without the necessity for the direction of gaze falling on the text. Three information unit text messages (what, where, and action) suggested for CMS messages by Dudek are probably too complex to be read quickly with near-peripheral vision.(3,5) However, the new technology CMSs (see the next section, New CMS Capabilities) are capable of displaying symbols that constitute a single item of information and might be recognizable in near-peripheral vision. Also, drivers might be able to detect single high-priority words, such as “ACCIDENT,” in near-peripheral vision.

The use of percent of time gazing at the forward roadway (i.e., percent road center (PRC)) has been proposed as an inverse measure of driver visual distraction to in-vehicle devices.(16) PRC would typically not be appropriate for assessment of distraction caused by CMS content because at least portions of the signs can be read without taking the driver’s gaze away from the forward roadway, which these authors defined as a 16-degree cone about the forward path. As noted in the previous paragraph, a cone that large would include obstacles on the road and an overhead CMS for at least 500 ft (152 m) of travel distance.

The standard deviation of gaze angle (SDG) (computed as the sum of the square root of x2 plus y2, where x is the difference between the previous horizontal gaze angle and the current gaze angle, and y is the analogous difference in the vertical plane) is reported to be sensitive to cognitive workload (mental distraction not caused by a competing visual task).(12) Therefore, SDG might be an appropriate indicator of distraction by CMS content if that content increases driver cognitive distraction. Victor et al. found that cognitive distraction decreased the standard deviation in gaze angle compared with a no distraction baseline.(16) However, they could not show that SDG varied with the other levels of cognitive distraction they employed (i.e., levels of distraction other than the no distraction baseline). They speculated that the easiest cognitive distraction task maximized workload so that more difficult tasks could not show additional decrements. The distracter task they used was to count the number of times a target tone occurred within a string of 15 tones. The difficulty was manipulated by varying the number of different target frequencies (two, three, or four) for which the participant had to listen.

Lane tracking, whether measured by steering entropy or lane deviations, would also not be a suitable measure of attention devoted to CMS content for the following reasons: (1) although lane tracking has been found to be sensitive to visual distraction by in-vehicle devices, gazes at CMSs would still allow monitoring of lane position with near peripheral vision, and (2) lane tracking has not been found to be sensitive to degradation from cognitive distraction.(12)

Reaction time to traffic events has been shown to be degraded by cognitive distraction as defined by increased mental workload. Thus, to the extent that CMS content is suspected of increasing mental effort, reaction time to events in the roadway might be used to test the presence of this distraction effect.

Time headway appears to be a weak predictor of distraction.(17) In an extensive series of on-road and driving study tests, there was some suggestion that visual distraction (attention to an in-vehicle device) results in slower speeds and longer headways, whereas cognitive distraction results in somewhat shorter headway in some studies, but not others.

New CMS Capabilities

Currently, CMS vendors are marketing ITS grade (National Transportation Communications for ITS Protocol (NTCIP) and National Electrical Manufacturers Association (NEMA) compliant) light-emitting diode (LED) CMSs for freeways with pixels as small as 0.47 inches (12 mm). Each pixel of a full-color CMS is made up of red, green, and blue LEDs. Such signs can display color images with symbols that closely match the MUTCD color specifications.(18) Although the ITS grade signs cannot display full-motion video or rapidly refreshed animations, commercial LED signs (often supplied by the same vendors that market ITS signs) can.

Current CMS Guidelines and Knowledge Gaps

This section provides an overview of the current state of knowledge concerning the human factors aspects of CMS content.

The following distraction effects are discussed in this section:

Attraction

The idea that distraction can be the result of an involuntary capture of attention is related to the orienting response in the classical conditioning paradigm. A bright light that suddenly and unexpectedly illuminates may cause an orienting response—a response in which the individual involuntarily shifts attention and gaze to the stimulus. Whether a CMS could cause an orienting response is questionable. Orienting responses are easily extinguished.(19) That is, if exposure to the stimulus is repeated and the stimulus is not related to current priority goals—in this case driving—then the orienting response to the stimulus will stop occurring. Thus, even if the sudden onset of a CMS image or message would cause an orienting response on first occurrence, drivers would soon learn not to orient to such changes unless those changes reliably predicted an event that has priority over watching the road and monitoring the vehicle state.

Information Overload

The construct of information overload was aptly described in National Cooperative Highway Research Program (NCHRP) Report 488:(20)

Information overload might occur if a driver perceives the need for the information on a TCD but extraction of the information requires more attention resources than the driver can devote to the message without risking degradation of driving performance.

Unintended Driver Reactions

Information overload may be possible when drivers either have to gaze at a CMS too long, or drivers have to devote too much attention (mental effort) to extracting meaning from a CMS. Two studies that appear to provide direct evidence of the distracting effect of information overload from CMS messages are cited here.

Erke, Sageberg, and Hagman conducted a study of driver responses to route diversion messages of CMS in Oslo, Norway.(21) At one site, the messages displayed indicated that a tunnel ahead was closed and suggested a ring road diversion. In fact, the tunnel was not closed. They found an increase of about 25 percent in the proportion of drivers exiting at the suggested ring road. Virtually all traffic left the freeway before reaching the tunnel. Thus, it appeared that the messages were nearly 100-percent effective in communicating that the tunnel was closed. However, there were unintended consequences. Mean speed was reduced in the proximity of the sign by about 5 mi/h (8 km/h). There was also a significant increase in speed variance and a significant increase in the number of brake light illuminations within reading distance of the sign. Video recordings suggested that part of the decrease in speed and braking events resulted when the lead vehicle in a platoon braked thereby requiring braking by closely following vehicles. Unfortunately, the authors did not measure legibility distance nor do they report the size of the text on the signs, which used amber LEDs. The text messages on the signs consisted of four lines but only three units of information (location, event, and action). The authors imply that the signs were legible from about 500 ft (150 m). At the posted speed of 50 mi/h (80 km/h), drivers would have had about 5.5 s to read the sign, or slightly less than the 2 s per information unit suggested by Dudek.(3) The Erke et al. study demonstrates that CMS can adversely affect driving performance and potentially affect safety. However, because the signs in this study appear to fall short in legibility distance and perhaps because the messages had more lines than are typical, it should not be concluded that these effects would be observed with messages that fully comply with the guidance provided by Dudek.

A similar slowing for four-line CMSs was reported for a simulator study that primarily focused on the effect of bilingual CMS messages on driver performance.(22) In that study, drivers slowed for four-line message signs, whether monolingual or bilingual, but did not slow for one- or two‑line messages. A significant reduction in time headway was also observed only for four-line message signs when the vehicle ahead varied its speed on approach to the signs. For bilingual four-line messages, the investigators tried to mitigate the performance changes by placing a blank line between the English and Welsh messages (two lines each). This mitigation strategy had no effect. However, the number of trials in which the blank line was present was small, and participants were not told ahead of time what the line indicated (English above, Welsh below). Thus, in locations where bilingual CMSs may be desirable in the United States (e.g., States that border Mexico and Quebec), strategies such as visual separation of message lines might still be effective if drivers know what the visual separation means and also have more than minimal exposure to the visual separation. Because Welsh, English, Spanish, and French use nearly identical alphabets, drivers need to process all the lines of a bilingual message to extract the part of the message that is meaningful to them. It is reported that when the characters are distinctly different (e.g., Japanese and English or Greek and English), four-line bilingual signs do not hinder driver performance.(22)

The studies described in this section suggest that speed and headway changes in the vicinity of a CMS are potentially useful driving performance measures for assessing the distraction effect of CMSs because erratic control of speed and headway could indicate a diminution of attention to the driving task. Note that if a driver slows to allow more time to read a CMS, the headway of the driver who slows will increase, but that will cause the headway of following drivers to decrease.

CMS Properties

This section reviews some of the CMS properties that, depending on implementation and operational variations, may be related to the human factors that may result in reduced or increased driver distraction.

Pixel Size (Dot Pitch)

Currently, NTCIP- and NEMA-compliant, full-matrix, full-color CMSs for freeway implementations have pixel sizes that range from 0.79 to 2.36 inches (20 to 60mm).(23,24) Each pixel consists of red, green, and blue LEDs. In some cases, the two red LEDs are used in combination with one blue and one green. Although the pixels, even those of 0.79inches (2 cm) in diameter, would seem to be large, at the viewing distances of 100 to 1,000ft (30.5 to 300.5m), the individual pixels can appear to form fairly smooth continuous lines and curves.

Font Issues

Previous research on the legibility of text on CMSs does not necessarily apply to CMSs with 0.79- or 1.33-inch (20- or 34-mm) CMS displays. Early research found that the number of matrix elements did not affect legibility distance.(25) Although Garvey and Mace tested a font that consisted of 0.79-inch (20-mm) pixels, they did not attempt to replicate FHWA fonts, which can be approximated with 0.79-inch (20-mm) pixels and 18-inch (46-cm) letters. They also did not test uppercase/lowercase lettering with 0.79-inch (20-mm) pixels.

Much of Europe is in the process of adopting the TERN font for CMS.(26) The TERN font is used on conventional European traffic signs, so in that sense, it is comparable to the FHWA standardfonts.

Garvey, Pietrucha, and Meeker reported that recognition distance (i.e., the distance at which destination names could be distinguished from each other) was greater when uppercase/ lowercase lettering (i.e., first letter capitalized and the remaining letters in lowercase) than all uppercase lettering.(27) Their stimuli were all positive-contrast white letters on a green (guide sign) background. Garvey et al. suggest that the uppercase/lowercase benefit derives from participants’ ability to distinguish word shapes from the configuration of risers and descenders in uppercase/lowercase text. In their study, participants were given a destination name to look for, and the participants responded by indicating whether the target destination was on the first, second, or third line of a sign. With no expectation regarding what words to anticipate, shape was of less benefit. Indeed, the investigators found that legibility distance did not vary between uppercase/lowercase and all uppercase destination names when the participants had no expectation of what destinations would be shown on the sign. Apparently on the basis of this finding, the 2009 MUTCD was amended to specify street name and destination names be in uppercase/lowercase font. Other text on guide signs is to remain all upper case.(2) This study’s research team was unable to identify research that examines whether case has an effect on recognition or legibility distance on high-resolution CMS displays. At least for positive contrast, similar effects may be true for CMSs. In tests with negative contrast text, Holick, Chrysler, and Park found no advantage for uppercase/lowercase text over upper case test.(28) However, this seemingly contrary finding may be because participants in the study had no expectation regarding the message content. For CMSs with a limited vocabulary and where location names are used, uppercase/lowercase text may provide a recognition distance benefit. This hypothesis is testable.

Symbols

McDougall, Tyrer, and Folkard reported search times for icons are faster when the icons possess a quality they refer to as “objectness.”(29) Quickly recognized icons are not necessarily the least complex (although complexity may affect legibility). Thus the I-10 route shield may be reacted to more quickly than the I-10 text if the text is treated perceptually as two objects and the route shield as one. This prediction is easily testable.

Icons that have good legibility distance, that are recognized in context by most drivers, and that are familiar are likely to speed sign comprehension. The only study this research could identify that appears to confirm this conjecture was difficult to evaluate. Wang, Collyer, and Clark reported that recognition of CMS messages with icons was slightly faster than text-only messages.(30) However, the generalizability of their findings is doubtful. Participants had to decide which of two messages was being displayed. In this situation, recognition of either the text or the icon was sufficient to support a response. Icons were used to communicate incident type (e.g., crash, congestion, and slippery wet pavement), and text provided the location and recommended action. Only the icon or first line of all text messages needed to be recognized. Had participants needed to read the entire text or icon plus text message, the results may have been different. Also, there were hints in the data of speed/accuracy tradeoffs. Although the correlation between speed and accuracy was not significant, accuracy was not analyzed as a dependent measure. Thus, the possibility that the use of familiar icons or symbols could speed sign comprehension or increase legibility distance has not been fully and unambiguously evaluated.

Motion

Motion or animation and flashing are not allowed by the MUTCD. Lane closure portable arrow boards, which are included in the MUTCD, use a sequence of chevrons to imply motion; however, these boards do not create true apparent motion and do not qualify as animation (see section 6F.61 of the MUTCD).(2) These boards are currently only approved for temporary lane closures in conjunction with other TCDs such as channelizing devices and static signs. With the advent of active traffic management, which includes lane closures using permanent overhead CMS, the evaluation of effectiveness of similar sequenced symbols seems warranted.

Luminance

The MUTCD does not specify the required luminance for CMSs other than dictating that it meets industry standards for day and night-time illumination.(2) The current industry standard is NEMA TS 4.(23) The MUTCD specifies that the luminance automatically adjust to ambient light and recommends that the contrast ratio between lit and unlit portions of the sign remains between 8and 12 (see section 2L.03, paragraphs 10 and 11 of the MUTCD).(2)

Contrast

As indicated in the previous paragraph, it is recommended that the contrast ratio between lit and unlit elements of CMSs be between 8 and 12.

The contrast ratio C is defined by the formula shown in figure 1.

Figure 1. Formula. Contrast ratio.

Figure 1. Formula. Contrast ratio.

Where:

Lon = luminance of an area of the display with pixels on.

Loff = luminance of the same area with pixels off.

Sun position is an important element of CMS legibility. When the Sun is shining from behind the driver onto the face of the sign, contrast can be considerably reduced. The amount of reduction depends on the maintenance of the sign and on the design of the sign face. NEMA TS 4 specifies the test procedure for verifying standard conformance when the Sun is 10 degrees above a perpendicular to the sign face.(23) When the Sun is shining from behind the sign into the driver’s face, apparent contrast will depend on the amount of shielding around the sign. Standards for shielding the sign or illuminating sign elements in this adverse lighting condition were not identified.

Messaging

The most complete guide to CMS message construction is probably the guide prepared for the New Jersey Department of Transportation.(5) That document provides guidance on message structure, vocabulary, length, flashing, and paging of messages. Although it provides thorough and apparently sound advice for CMS operators, the empirical foundations for that advice are not cited. These foundations may be found in earlier publications, many of which are cited in the following sections.

Flashing

Currently, the MUTCD does not allow flashing any part of a message on a CMS (see section 2L.05, paragraph 5).(2) Numerous studies have looked at the effects of flashing messages on CMS. Flashing is generally intended to increase the conspicuity or attention capturing of important CMS messages.(31) However, few evaluations examine the effects of flashing on attention capture. One study that did look at the effect of flashing on conspicuity was reported by Charlton.(32) In that study, the effects on conspicuity were mixed. A flashing school warning sign increased search conspicuity and memory for that warning compared with a static school warning sign. A flashing road work ahead sign also proved more conspicuous than a static road work ahead sign of the same size, but was not more conspicuous or memorable than an oversized static version of the same sign. Charlton concluded that the effectiveness of flashing for attracting attention and improving memory for CMS messages may vary with the message. However, the Charlton study looked at only a small subset of CMS symbols and no text messages. The overall effectiveness of flashing as a conspicuity enhancement deserves further research.

A number of studies have shown that flashing may make CMS text messages harder to read. These studies examined reading time and comprehension for flashing messages and did not weigh the benefits of faster reading time or better comprehension against the potential for more drivers to attend to the sign or the likelihood that the sign would affect driver behavior in the desired manner.(33–35) One study in which messages were presented on a laptop computer found that reading times were increased by about 20 percent if any part of a message was flashed.(34) This finding seems reasonable because the amount of time the message is exposed is decreased when it flashes. Nonetheless, the Dudek 2006 study failed to replicate the reading time effect in a driving simulator.(33) Comprehension was not affected when an entire three-line message was flashed.(34) However, recall of the third line of simulated CMS messages was depressed when only the first line of the message flashed.

No research was identified that looks specifically at whether flashing on a CMS distracts drivers in a manner that would adversely affect safety. Dudek et al. did not find effects on driving performance in a simulator as a result of flashing.(34) In that study, the roadway was straight, and drivers had to monitor the forward roadway to maintain a safe following distance behind a simulated vehicle that braked unpredictably.

Flashing is used by some agencies to increase the conspicuity and apparent importance of CMS messages. There is a dearth of research showing that flashing accomplishes its intended purpose and little research that shows negative effects on reading time or comprehension that would outweigh the putative conspicuity benefits. Although the current body of research does not support elimination of the ban on flashing, neither is there evidence that either operational effectiveness or safety is adversely affected by the ban. Research in this area might clarify these issues.

Vocabulary

The appropriate vocabulary and syntax for text-based messages was one of the earliest areas of human factors CMS research and analysis. (See references 36 through 40.) More recently, research has focused on methods of presenting travel times on CMS.(41,42) The research in this area seems adequate to meet user needs for permanent full-matrix color CMSs.

Message Length

CMS technology (e.g., the number of characters that can be displayed) and the amount of time CMS messages are legible to drivers constrain the amount of information that can be displayed. When approaching drivers are traveling at freeway speeds, CMS text with 18-inch (46-cm) characters is legible for a maximum of about 8 s. The MUTCD specifies 18 inches (46 cm) as the minimum letter height for CMSs on freeways (see section 2L.04, paragraph 6 of the MUTCD).(2) Because CMS legibility distance has been found not to increase much beyond that afforded by an 18-inch (46‑cm) letter height, this height has tended to become the standard height. Obstructions, such as other vehicles, horizontal of vertical curves, and adverse atmospheric conditions (e.g., rain), can further reduce the amount of time that drivers are exposed to legible messages. Furthermore, early research in driver glance behavior showed that drivers do not look at signs in single long glances.(13) Rather, drivers make multiple short glances interspersed with glances to the roadway and other vehicles. Thus, for a static sign that is legible for 8 s, the total time a driver may glance to areas where the sign text can be read will sum to only about 2 s on average but may be as much as 6 s.(13) Dudek has estimated that each “unit” of information on a CMS can be read in about 2 s.(3) He recommends that CMS messages be limited to three or at most four units of information. Dudek defines a unit of information as the answer to a simple question. For CMS these questions are usually what (e.g., crash), where (e.g., ahead), and what action should be taken (e.g., use frontage road).

In the section on unintended driver responses, this report noted that some studies with messages that probably violate the Dudek guidance on message length resulted in drivers slowing or braking. There is also some evidence that at least some drivers slow in response to CMS messaging even when the messages are properly formatted.(43) It is unlikely that improvements in CMS technology could justify text-only messages longer than those currently recommended.

Abbreviations

No research was identified that showed that abbreviations that are understood by more than 85percent of drivers take longer to comprehend than the complete words that would be used in their stead. Huchingson and Dudek and Hustad and Dudek reported many of the abbreviations specific to portable CMSs are almost universally understood when shown with a prompt—a word with which the abbreviation would always appear (e.g., New Jersey TRNPK).(44,45) To date, the study of highway sign abbreviations has only considered comprehension. The effect of abbreviations on reading and comprehension time and effort should be considered in addition to simple comprehension measures because the time available for drivers to view CMS messages can be short.

Phasing

Of all the currently allowed approaches to displaying messages on CMSs, phasing is probably the most problematic in terms of the potential for driver distraction. Phasing, or paging, breaks a message into two or more parts, with each part displayed in sequence on the same sign. Phasing is used to overcome space limitations. The MUTCD limits phasing to two parts, with each part having no more than three lines (see section 2L.05, paragraph 4).(2) The minimum duration of a phase is 1 s per word or 2 s per information unit. The maximum cycle time is 8 s, which would mean that at freeway speeds, a driver would be exposed to only one cycle during the time that the message is legible. The MUTCD also requires that the intelligibility of the message not depend on the order in which the two phases are read.

The distraction potential of phasing is high because drivers typically make multiple short glances to signs rather than single long glances. The first glance could come at any time during either phase. If more than one glance is required to extract the information from a phase, there is no guarantee that the same phase will be present when the driver returns for a second glance. If the second glance arrives during a new phase, the driver will need to reorient attention to the new message phase. Once oriented, a third glance may be necessary to begin processing the second phase. A fourth glance might be needed to complete processing of the second phase, but this glance could potentially land not on the second but on the first phase, which would again call for a reorientation of attention. All this would be done while the driver continues to monitor the roadway, so that timing of glances to the sign would depend on what the surrounding traffic is doing. Furthermore, the MUTCD allows for a blank interval between phases of up to 0.3 s. Research has suggested that readability, measured by recall, is improved by the insertion of a 0.3s blank interval between phases. A blank phase may prevent masking of one message by another, or it may just allow drivers time to prepare to attend to a new phase, e.g., shift attention to the upper left portion of the display. However, a driver’s glance might arrive during a blank interval, which could either cause a driver to linger on the display until a message appears or return attention to the road and delay reading the sign. In any case, phasing increases the effort and time required to read a CMS and thus will increase the probability that attention will be on the sign when a safety-critical event occurs.

No previous studies were identified that examined the effects of CMS phasing on driver behavior in the field or employed eye tracking to quantify how drivers manage CMS messaging with the other visual requirements of the driving task. The effects of phasing on reading time are reported in chapter 4 of this report.

Supplemental Guide Signs

The MUTCD describes the use of supplemental guide signs (see figure 2) as follows:(2)

Figure 2. Photo. A series of four consecutive supplemental guide signs at a Virginia interchange

Figure 2. Photo. A series of four consecutive supplemental guide signs at a Virginia interchange.

Supplemental guide signs are approved for guiding unfamiliar travelers to certain types of destinations. Destinations specifically approved for supplemental signing are major colleges and universities, large military bases, major event facilities (e.g., sports arenas and stadiums), State and national parks, monuments, and major recreational areas.(46) For each of these destination types, the American Association of State Highway and Transportation Officials (AASHTO) has provided specific minimum traffic generation criteria. Among the destinations for which supplemental guide signs are deemed inappropriate are businesses, churches, government research centers, driver’s license centers, any schools not qualified as major colleges or universities, and medical facilities.

The MUTCD recommends that no more than one supplemental guide sign be installed per interchange and not more than two destinations be listed on a supplemental guide sign. AASHTO also provides specific maximum distance from interchange criteria that should be met. The AASHTO guidelines are incorporated into the MUTCD by reference (see section 2E.35).(2) If a supplemental guide sign is used, it is to be placed approximately midway between advance guide signs for an interchange where two advance guide signs are used or 800 ft (244 m) after a lone advance guide sign.

The recommended limit of one supplemental guide sign per interchange and the spacing of supplemental guide signs at least 800 ft (244 m) from other signs is intended to avoid possible driver information overload. However, there appears to be no direct empirical evidence that this amount of signing will mitigate information overload or that more signing than is currently permitted would result in information overload. Therefore a priority for the data collection portion of this project was to measure driver performance as a function of the frequency and spacing of guide signs, specific-service signs, and permanent overhead CMSs. In conducting this research, it was assumed that the driver is unfamiliar with the roadway and adjacent facilities and is seeking specific destinations.

Approach

The objectives were addressed through a series of studies. Each of these studies is the subject of a separate chapter.

Chapter 2 briefly describes a laboratory study to obtain psychophysical measurements with a state-of-the-art NTCIP- and NEMA-compliant CMS. This effort included asking observers to judge the similarity of symbols and messages displayed on an actual CMS with variations of the same symbols and messages on a liquid crystal display (LCD). The purpose of collecting the similarity comparison judgments was to assess the requirements for conducting CMS messaging research with simulated CMS displays.

Chapter 3 examines the legibility distance of CMS messages in relation to letter height and the observer’s visual acuity.

Chapter 4 describes a field test with the state-of-the-art CMS in which drivers read CMS messages in a controlled driving environment. Reading time, reading distance, and eye glance behavior were recorded as drivers approached the CMS while maneuvering an instrumented vehicle at a constant speed. The results are reported for phased messages, flashing messages, messages with abbreviations, messages of varying length, and text versus symbol messages.

Chapter 5 describes a driving simulator study that examined the following issues:

Chapter 6 describes a study that is very similar to that of chapter 4. However, this chapter examines whether the content on the CMS changes the probability that the driver will detect a hazard in the roadway.

Chapter 7 examines how the frequency and spacing of freeway guide signs affects driver behavior and the ability to detect guide sign information.

 

 

Federal Highway Administration | 1200 New Jersey Avenue, SE | Washington, DC 20590 | 202-366-4000
Turner-Fairbank Highway Research Center | 6300 Georgetown Pike | McLean, VA | 22101