DIRECT
`Operational Field Test Evaluation
`Human Factors
`
`Submitted to:
`Michigan Department of Transportation
`Traffic & Safety Division
`Transportation Systems Section
`425 West Ottawa Street
`P.O. Box 30050
`Lansing, MI 48909
`
`August 1998
`Contract No. 94-1519 DAB
`
`Prepared by:
`University of Michigan
`EECS Department
`ITS Research Laboratory
`
`Jill Fleming
`Paul Green
`Stewart Katz
`
`200 Engineering Programs Building
`2609 Draper Drive
`Ann Arbor, MI 48109-2 101
`Phone: (734) 764-4333
`FAX: (734) 763-1674
`
`EECS-ITS LAB-DT98-003
`
`
`
`FORD EX. 1022
`
`

`

`Technical Report UMTRI-98-22
`(also listed as EECS-ITS LAB-DT98-003)
`
`June, 1998
`
`Driver Performance and
`Memory for Traffic Messages:
`Effects of the Number of Messages,
`Audio Quality, and Relevance
`
`Jill Fleming
`Paul Green
`Stewart Katz
`
`
`
`
`/
`
`
`I
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`FORD EX. 1022
`
`

`

`2 Government Accession No
`
`3 Recipient s Catalog No
`
`Technical Report Documentation Page
`
`5 Repor Date
`June, 1998
`6 PerformIng Organization Code
`none
`8 Performing Organlzation Report No
`UMTRI-98-22
`10 Work Unit no (TRAIS)
`
`1 Report No
`UMTRI-98-22
`4 Title and SubtItle
`Driver Performance and Memory for
`Traffic Messanes: Effects of the Number of
`Messages, Audio Quality, and Relevance
`7 Auth or(s)
`Jill Fleming Paul Green, Stewart Katz
`9 PerformIng Organization Name and Address
`The University of Michigan
`Transportation Research Institute (UMTRI)
`2901 Baxter Rd, Ann Arbor, Michigan 48109-2150
`12. Sponsonng Agency Name and Address
`Michigan Department of Transportation
`Traffic and Safety Division
`Transportation Systems Section
`425 W. Ottawa (P.O. Box 30050), Lansing, MI 48909
`Attention: Dr. Kunwar Rajendra
`15 Supplementary Notes
`also identified as report # EECS-ITS LAB-DT98-003
`Abstract
`In this experiment, 32 licensed drivers (16 young, 16 old) drove on an expressway.
`On each trial (96 per subject), 1 to 3 traffic messages containing 6 to 14 items were
`presented. (“l-94 eastbound at Southfield freeway, continuing construction, right lane
`blocked, 3 mile backup.“).Imagining they were driving from Ann Arbor to Detroit on
`l-94, subjects identified messages relevant to that route and recalled them. Messages
`were either of good or poor audio quality (to simulate poor reception).
`Drivers familiar with the route correctly recognized about 85 % of the relevant
`messages. Typically 4 items were recalled regardless of message length, with the
`road and crossroad being most common. Drivers recalled the direction to which the
`message pertained (e.g., l-94 east) only 39 percent of the time. Drivers believed the
`traffic information system was safe and useful to listen to while driving (approximately
`9 on a lo-point scale) and would pay $177 on average for one, though most subjects
`were not willing to pay anything.
`
`11 Contract of Granl No
`
`13 Type of Report and Period Covered
`final
`
`14 Sponsoring Agency Code
`
`Of the message characteristics, message audio quality had the largest adverse
`impact on performance (increasing speed variance). Differences were also found
`between (1) driving, (2) driving while listening to messages (slight increases in speed
`variance), and (3) driving while speaking with the experimenter (further increases in
`speed variance). This is in contrast to claims that dealing with auditory information
`while driving has no impact on driving workload and is not a safety concern.
`17 Key Words
` 18. Dlstnbutlon Statement
`ITS, ‘human factors, ergonomics,
`No restrictions, This document is
`driving, usability, safety,
`available to the public through the
`traffic information, auditory interfaces
`National Technical Information Service,
`Springfield, Virginia 22161
` 20. Secunty Classlf. (of this page)
`1 21. No of pages
`none
` 87
`Reproduction of completed page authorized
`
`19. Security Classlf. (of thts report)
`none
`Form DOT F 1700 7 (8-72)
`
` 22. Price
`
`FORD EX. 1022
`
`

`

`=—
`
`Driver Performance and Memory of Traffic
`Messages: Effects of the Number of Messages,
`Audio Quality, and Relevance
`Boe
`UMTRI Technical Report 98-22
`University of Michigan,
`Jill Fleming, Paul Green, and Stew Katz
`Ann Arbor, Michigan, USA
`ER issues
`1. How doesrecall vary with message content (number of messages, terms, and
`relevant messages), message quality, and driver differences (age and sex)?
`
`2. How do driver performance (speed, headway,lateral postition) and control inputs
`(throttle, steering) vary with the message and driver characteristics?
`
`3. How easy and safe to use do drivers rate auditory traffic information systemsrelative
`to other in-vehicle tasks?
`
`4. Whatis the rated usefulnessoftraffic information systems and of each information
`element? Would drivers use such systems? How much would they pay for them?
`SET
`wartee mle
`we
`~
`wee ee eee RLa
`
`Described
`Route
`
`Ford Rd.
`
`wcngat >
`
`How many
`messages?
`
`Expernmenter:
`“Wasthat relevant to
`your route?"
`
`Expenmenter:
`"Were any of those
`relevant to your route?"
`
`|Male|
`
`|4|
`
`in worksheet
`
`Iterations
`
`Message Relevance
`Relevant
`Irrelevant
`
`Experimenter:
`Expenmenter:
`"Repeat as much
`Which
`information as you can
`Good
`Good
`wonessage(s)
`rememberfom the
`messages
`ptdt8h8
`Record information
`pepe
`
`FORDEX. 1022
`
`FORD EX. 1022
`
`

`

`Sew tbe esos +
`
`c
`art 38
`& £
`n& 3
`OBe 6
`® ©
`2%
`A's 3.4
`BO
`@
`=
`
`32
`
`= 110
`—
`~o
`g 2
`3S
`o D
`ra
`c;D
`B og 102
`= = 100
`x
`Q 98
`
`Young
`
`Numberof
`Messages
`~O- |
`2
`a 3
`
`Baseline Message Response
`Segmentwithin Trial
`
`Baseline Message Response
`Segment within Trial
`
`Relevant
`
`El RESULTS Pp
`
`2 Shape expected from||¢ =
`
`
`
`25 serial position curves||= § Irrelevant
`) =
`a2
`os
`— 2
`a
`os
`2 ®
`<2
`©
`o5
`=P
`2 oO
`c
`D 6
`oo
`af
`2 c
`8 B[~B”
`53 ES 38 8 8
`oO
`5 § 8 3
`££
`&§
`BS ES
`5
`$§ £5 8
`8
`&
`8 = 8
`5
`$e &
`eB
`5
`|
`2
`¢
`*
`8
`=!
`3
`=
`°
`
`Good
`Poo
`Message Quality
`Amount Willing Numberof
`toPay(U.S.$)
`Subjects
`0-50
`12
`70 - 100
`6
`200 - 500
`6
`
`SASSetoFgDeraSt7TEREEESSeISSSESSELESERERSASETSacSEaePRBeaSORESCNAeReMCTIRE2Fy,PBISESEyREETSTee8
`
`.
`.
`Serial Location
`of Term
`
`S
`z
`
`CONCLUSIONS
`
`1. Poor audio quality combined with lack of relevance, more than one message, and
`age led to reduced recognition of the messages. Drivers recalled less than half of the
`information presented. Approximately 4 terms from eachtraffic message were
`recalled, regardless of the amount presented.
`
`2. Poor audio quality, lack of relevance, more than one message, and response tasks
`led to poorer driving performance.
`
`3. Drivers generally felt that the system was safe for them to use, but did not believe
`that it was safe for inexperienced drivers.
`
`4. Subjects felt that the information would be useful when driving in a familiar area, but
`not in an unfamiliar area. On average,drivers were willing to pay $117 (U.S.) forthis
`type of system. However, the most commonresponse was $0.
`
`iv
`
`FORDEX. 1022
`
`FORD EX. 1022
`
`€
`

`

`PREFACE
`
`This report describes an on-the-road experiment conducted as part of the DIRECT
`(Driver Information Radio using Experimental Communication Technologies) Project,
`an operational field test of various low-cost traffic information systems sponsored by
`the Michigan Department of Transportation (MDOT) using funds provided by the U.S.
`Department of Transportation, Federal Highway Administration (FHWA). In this project
`4 mechanisms for presenting traffic information to drivers were evaluated: (1) Low
`Power Highway Advisory Radio (LPHAR), (2) Automatic Highway Advisory Radio
`(AHAR), (3) cellular call in, and (4) Radio Broadcast Data System (RBDS).
`
`LPHAR is a descendant of Highway Advisory Radio (HAR), a system that utilizes
`roadside signs with lights. A flashing light indicates when drivers should tune to a
`local station for traffic information. Unlike variable message signs, the message is not
`limited by the size of the sign or the vehicle speed which determines the time available
`to read the message. The visual distraction of reading the sign is also eliminated.
`LPHAR, a radio signal, has a range of 1.0 to 1.8 miles, localizing the message and,
`therefore, allowing for multiple messages in a region.
`
`AHAR is similar to HAR (from the driver’s perspective) except that the information is
`presented to the driver automatically, interrupting (if the driver so chooses) any
`broadcast the driver is listening to at the moment. However, the equipment used can
`be quite different from that of HAR.
`
`In cellular call in, the driver calls a particular phone number for traffic information.
`Options considered at various times included individual phone numbers for each road
`and a single phone number after which the driver entered the route number.
`Eventually, to provide more locally specific information, other data could also be
`needed (e.g., nearest exit).
`
`In the RBDS system, a system that originated in Europe, traffic information is presented
`on special channels that can be received by modified automotive radios. Also known
`as RDS-TMC (Radio Data System - Traffic Management Channel), a display with text
`messages may also be provided. The RDS-TMC system provides for interrupting
`ongoing broadcasts. To avoid driver overload and the presentation of all possible
`area traffic messages, a filter indicating the current route must be programmed by the
`driver. During the planning of this program, the authors were unaware of the driver
`task of entering the filter, a task that could prove challenging to many drivers and
`deserves examination.
`
`To examine the merits of these systems, 5 aspects pertaining to the implementation
`were investigated. In the natural use study (1), 150 drivers were loaned vehicles fitted
`with these systems for 2 weeks or 2 months for their own use (Reed, Hanafik,
`Richeson, and Underwood, 1998). Survey data concerning their usage and opinions
`were obtained. The simulation and modeling effort (2) examined the improvements in
`traffic flow in the Detroit area as a function of various levels of market penetration of
`these systems (Underwood, Juna, Gurusamy, Hadj-Alouane, and Hadj-Alouane,
`1998). Technical performance and costs were examined (3) to determine how well
`these systems functioned (signal quality, message accuracy, etc.) and to estimate
`
`V
`
`FORD EX. 1022
`
`

`

`production costs (Ristenbatt and Shahine, 1998). Pan of this effort included the
`collection of intelligibility data. Institutional and organization issues (4) were
`considered as part of the broader project view (Richeson, Underwood, and Waldman,
`1998).
`
`This particular report describes the final area (5), the human factors research
`conducted as part of the DIRECT project. Of particular interest was the safety and
`performance impacts of using the 4 systems of interest. The initial discussions of this
`project centered on providing a broad human factors evaluation of all systems.
`However, given the limited funds available, such an approach would have been
`superficial, adding little to the scientific literature.
`
`The approach taken was therefore to identify the safety and human factors issues, and
`target those that seemed most important and common to all systems. Also considered
`was the extent to which the research would provide new information, not information
`that would duplicate the literature. Issues of concern were reading the RDS-TMC
`display, pressing buttons to retrieve messages, listening and responding to messages,
`and dialing the cellular phone. Given the limited number of characters on the display
`when the system was initially being discussed and uncertainty about how it would be
`implemented, RDS-TMC display issues were set aside for future efforts. Further, most
`systems only required a single button press to retrieve information, a task that was not
`thought to pose much risk to drivers. Set aside for future investigation was retrieval of
`information (the keying task) for a cell phone. The initial design only required calling a
`particular number, a task examined in the literature (Goodman, Bents, Tijerina,
`Wierwille, Lerner, and Benel, 1997). However, later implementation may require
`navigation through menus, a task deserving examination.
`
`The focus, therefore, was on tasks common to all interfaces, such as listening to traffic
`messages, determining if traffic messages were relevant, and attempting to recall
`those messages as a function of the amount of information presented. Eventually
`results from such efforts will include enhanced design guidelines for the presentation
`of auditory traffic information.
`
`The opinions, findings, and conclusions expressed in this publication are those of the
`authors and not necessarily those of the Michigan State Transportation Commission,
`the Michigan Department of Transportation, or the Federal Highway Administration.
`
`vi
`
`FORD EX. 1022
`
`

`

`INTRODUCTION .......................................................................................................................1
`Previous Research ........................................................................................................1
`Research Issues Investigated .......................................................................................6
`
`TEST PLAN...............................................................................................................................9
`Test Participants ............................................................................................................9
`Instrumented Car .........................................................................................................10
`Traffic Messages..........................................................................................................13
`Audio Compact Disc ....................................................................................................17
`Test Route....................................................................................................................17
`Test Activities and Their Sequence.............................................................................18
`
`RESULTS ................................................................................................................................21
`Recognition of Relevant Messages.............................................................................21
`Message Content Recall..............................................................................................23
`Ratings of Safety, Ease of Use, Usefulness, and Willingness to Pay........................28
`Driving Performance and Message Characteristics....................................................31
`
`CONCLUSIONS......................................................................................................................39
`How Well Did Drivers Recognize Relevant Messages? .............................................39
`How Well Did Drivers Recall Information from Traffic Messages?.............................39
`Did Drivers Believe It Was Safe for People to Listen
`to Traffic Messages While Driving?.......................................................................40
`How Useful Was the Traffic Information System?......................................................41
`How Useful Were the Traffic Information Elements?..................................................41
`How Much Were Drivers Willing to Pay for a Traffic System? ...................................41
`Did Use of the Traffic Information System Affect Driving Performance
`and What Were the Effects of Various Message Characteristics? .......................42
`What Should Be Done in Future Studies? ..................................................................44
`
`
`
`REFERENCES ........................................................................................................................47
`
`APPENDIX A - MATCHING TASK.........................................................................................49
`
`APPENDIX B - TEST VEHICLE LAYOUT.............................................................................51
`
`APPENDIX C – INSTRUCTIONS...........................................................................................53
`
`APPENDIX D - SAMPLE OF RECALL AND RECOGNITION
`WORKSHEET ...............................................................................................57
`
`APPENDIX E - SUBJECT SURVEY ......................................................................................59
`
`APPENDIX F - RECALL AS A FUNCTION OF MESSAGE LENGTH.................................61
`
`APPENDIX G - ANOVA OF DRIVING DATA........................................................................67
`
`FORD EX. 1022
`
`

`

`APPENDIX H - DETAILED DISCUSSION OF THE DRIVING DATA............................69
`Mean Speed .......................................................................................................69
`Standard Deviation of Speed ..............................................................................70
`Standard Deviation of Throttle ............................................................................72
`Mean Headway...................................................................................................73
`Standard Deviation of Headway..........................................................................74
`Standard Deviation of Lateral Position................................................................76
`Standard Deviation of Steering Wheel Angle ......................................................77
`
`FORD EX. 1022
`
`

`

`INTRODUCTION
`
`This project was conducted as part of the U.S. Intelligent Transportation Systems (ITS)
`program. The goal of this program is to improve the efficiency and speed with which
`goods and people are moved, to make transportation safer, and to make travel more
`enjoyable. This particular project concerns the movement of motor vehicles over
`public roads. To a significant degree, the movement of goods and people is
`hampered by congestion. Although congestion has been a problem both for
`expressways and city streets, the emphasis here is on expressways.
`
`One way to improve system efficiency is to provide drivers with better information
`about congestion, so that they might drive around congested areas or alter departure
`times. A variety of methods have been developed for that purpose. In the U.S., the
`most common method for people to obtain traffic information while driving is from traffic
`reports broadcasted by AM/FM stations, Although such information is generally
`complete, it may be dated (due to delays in updates of the radio stations by the police
`or traffic control centers, or because messages are presented periodically, e.g., every
`20 minutes). Further, broadcasts generally cover an entire metropolitan area, even
`though drivers are only interested in a small portion that applies to their route.
`
`In some cities variable message signs are popular. However, these signs can be
`distracting to read, and for long messages, a source of congestion rather than
`congestion relief.
`
`Previous Research
`
`Consequently, there has been considerable interest in audio-based systems that
`provide localized traffic information, especially systems that provide information about
`an entire trip so that alternatives can be considered at the beginning of a trip, not after
`one is caught in congestion.
`
`One of the alternatives considered was Highway Advisory Radio (HAR), a system
`developed by the FHWA in the early 1980’s (Turnage, 1980). When problems occur,
`drivers are advised by a flashing light on a sign to tune their radios to a particular
`frequency for further information.In fact, it was in conjunction with the development of
`HAR that virtually all of the research on understanding of auditory information while
`driving was conducted in the late 1970’s and early 1980’s. Other related work on
`auditory route guidance (e.g., the Back Seat Driver research at MIT, Davis, 1989;
`Davis and Schmandt, 1989) will not be covered here.
`
`The initial work on the presentation and retention of auditory messages while driving
`was conducted by Gatling of FHWA (Gatling, 1975, 1976, 1977). All of his studies
`followed the same basic format. (See Green, 1992 for a summary.) Subjects drove a
`car on a limited access highway while either tape-recorded messages were presented
`or slides were shown on a screen in the vehicle (simulating a head-up display (HUD)).
`Gatling’s performance measure was the percentage of subjects making a “route error,”
`that is, not recalling the entire message correctly. Variables manipulated included the
`modality of the information (auditory versus visual), the repetition of auditory
`
`1
`
`FORD EX. 1022
`
`

`

`messages, driver age (young versus old), and the number of items in the message.
`("Next right exit; for Boston; via route 213; 3 miles" would be a four-item message.)
`
`In the first of 4 experiments described by Gatling (1975), subjects heard messages
`containing 1 to 6 chunks of information, presented either once or twice. During a 5 to
`15 second delay crivers read aloud unrelated messages(e.g., “slow - automobile
`accidentin right lane") that interfered with rehearsal of the to-be-remembered
`message.
`
`As expected, there was noeffect of the delay on recall since the duration of the
`interfering task in the delay period wasfixed. Card, Moran, and Newell (1983) state
`that the half-life for working memory (middleman estimate) is 73 seconds for one
`chunk and 7 secondsfor three chunks. Using a 4-second interference period (to read
`the message), the predicted values are 96 % and 67 %correct, reasonably close to
`the 97 % and 46 % measured. Error rates were linearly related to the numberof units
`in the message, being 100 percent (no one recalled the entire message) for older
`drivers at 5 units/message and 100 percentfor youngerdrivers at 6 units. Presenting
`test messages a second time improved recall by about 15 %. Figure 1 shows some
`example results.
`
`Single Older
`
`Single Young
`
`O
`
`5
`
`OD
`
`_
`
`f
`
`OC
`
`=
`
`f
`-
`|
`-
`
`é
`:
`
`fms
`‘if
`#£
`/
`
`Double Young
`Double Older
`
`2
`
`5
`4
`3
`Numberof Units in Message
`
`6
`
`7
`
`100
`
`8041
`
`60
`
`40
`
`207
`
`0 1
`
`Error
`(%)
`
`Figure 1. Percent error versus message size, experiment 1 of Gatling (1975).
`Note: Single or double refers to presentation of the message onceortwice.
`
`Probing the drivers for just one piece of information (experiment 2) raised the levelat
`which 100 % errors occurred to 8 units/message. Repeating a message had the same
`effect as in the first experiment. Recognition was notidentical forall information
`
`2
`
`FORDEX. 1022
`
`FORD EX. 1022
`
`

`

`elements, being about 80-90 % for exits, streets, towns, distances, and turns, but only
`50 % for route numbers. The third experiment also showed problems with recall of
`numeric data while the fourth concerned the time to tune the radio to obtain a
`message.
`
`Gatling (1977) found that providing additional prose (“This is a warning message
`that...“) led to slightly better retention (by about 5 %) than more succinct messages.
`
`In the last experiment in this series, Gatling found that for messages with 2 route
`numbers or less, there were fewer errors with visual messages. For 3 or more routes,
`auditory messages led to fewer errors.
`
`The Gatling research comprises an interesting set of experiments concerning memory
`of route information. Notable is the use of on-the-road context to assess recall.
`Unfortunately, the nature of the interfering task was not precisely defined and the
`duration of the interfering task varied, so the results could be linked more closely to
`other research in the psychological literature.
`
`Subsequent human factors research was conducted by the Texas Transportation
`Institute (TTI) to develop design guidelines for HAR (Dudek, Huchingson, and
`Stockton, 1981; Dudek, Huchingson, and Brackett, 1983). In those experiments,
`subjects driving on an interstate highway were presented with a tape-recorded HAR
`message advising of an accident and a diversion route. Drivers then attempted to
`recall the message and drive the route. Four experiments were conducted. The first
`examined the number of information elements (exit street, direction of turn, etc.) and
`the language style (complete sentences/conversation; short form; staccato, e.g.,
`“overturned truck ahead“). The longer messages (10 versus 8 or 6 information
`elements) and less conversation formats led to more errors. In the second experiment,
`internal repetition (where the key elements were repeated as part of the message,
`“turn right on Kingman and take Kingman to Anderson”) were found to lead to fewer
`errors than for external repetition (where the message was given followed by “I repeat
`. . .’ In study 3, adding landmarks and information on the number of traffic lights helped
`people negotiate routes even though it lengthened messages. In study 4, unfamiliar
`drivers were found to make about the same number of driving errors when given turn
`information as did familiar drivers without turn information. This research led to the
`following guidelines:
`
`1. Although language style was not found to be critical, a terse message
`style was preferred by drivers. Unnecessary wordiness is inefficient in
`communicating messages in a HAR system.
`
`2. If unfamiliar drivers are diverted, the routes should not exceed 4 turns
`and 4 names, including the Interstate (8-unit problems).
`
`3. The description of the diversion route should be repeated at least
`once, either with internal or external redundancy or with both.
`
`4. Prominent landmarks may be mentioned in a HAR message whenever
`there is a risk the driver may not see the place to turn. The number of
`
`3
`
`FORD EX. 1022
`
`

`

`traffic lights is useful but should be avoided whenever any of the lights
`are flashing.
`
`5. When the driving population is known to be largely commuters or
`highly familiar with the area, the route description may be shortened by
`omitting turn directions. (Dudek, Huchingson, and Brackett, 1983, p. 9).
`
`Additional recommendations appear in Turnage (1981), p. 25-26.
`
`Audible messages:
`
`1 . . . . The length of a HAR message should be such that the motorist will
`hear it at least twice while within the HAR zone of coverage...
`
`4, Motorists remember names better than numbers. The greater the
`frequency of route numbers in a message, the greater the number of
`route errors made.
`
`5. Motorists retain cautionary messages better than informational
`messages.
`
`Visual signaling:
`
`1. It takes a motorist about 60 seconds after seeing the first advance
`visual HAR sign to turn on his radio and tune to the appropriate
`frequency. . .
`
`3. Motorists have been found to interpret the degree of urgency to sign
`messages as follows:
`
`greatest degree of urgency
`TRAFFIC ALERT
`moderately urgent
`TRAFFIC ADVISORY
`TRAFFIC INFORMATION least urgent
`
`It is very important that the motorist not be led to expect a message
`4.
`when the HAR station is not operating.
`
`As was suggested earlier, it is peculiar that all of these studies concern human short-
`term memory, but there has been little effort to connect these applied studies with the
`results from the psychological literature. For those unfamiliar with the literature, there
`are 3 forms of human memory linked in a serial fashion. The first form, perceptual
`store, is involved with the immediate readout of information. Information is stored
`physically and is generally lost if not attended to within just over a second or less.
`Depending on the sensory modality for which information is being stored, perceptual
`memory has a capacity of 5-17 items. This form of memory is operating when people
`glance at something, look elsewhere, and attempt to recall what was just seen. At the
`other end of the sequence is long-term memory. Information in long-term memory is
`usually stored semantically and its capacity is unlimited. This is usually the type of
`memory being invoked when people say they have memorized something such as
`
`4
`
`FORD EX. 1022
`
`

`

`their own name, the presidents of the United States, multiplication tables, the scent of
`a rose, the sound of a robin, or the taste of chocolate.
`
`Connecting those two forms of memory is short-term or working memory. Short-term
`memory has a capacity of about 7 chunks of information. Information is maintained in
`short-term memory by rehearsal and decays exponentially with time if not rehearsed.
`For example, after looking up a telephone number, one repeats the number to oneself
`“7 6 3, 3 7 9 5; 7 6 3, 3 7 9 5” to avoid,forgetting it. If the repetition process is
`interrupted (“What time is it? It is 3:15 p.m.“), the to-be-remembered information is
`often forgotten. Short-term memory along with some aspects of long-term memory are
`operating when drivers are asked to remember traffic information.
`
`Short-term memory recall and recognition probabilities are determined by a limited set
`of rules. These rules have implications for auditory traffic messages.
`
`1. Information is stored in chunks, units over which people group information. For
`example, the string “J M X” would be 3 chunks of information to most people (unless
`it was there initials or some other memorable item), while NBC would be 1 chunk.
`
`2. Generally, people can keep 7 chunks of information in memory, at least in laboratory
`situations. For highly reliable recall in real world situations, confine the to-be-
`remembered information to 4 chunks. This is consistent with the 4 turn-4 road
`results in the literature.
`
`3. Since information decays exponentially with time, the decay rate is often specified
`as a half-life, the time period over which half of the information initially available is
`lost. That time is just over 70 seconds for 1 chunk of information, 7 seconds for 3
`chunks.
`
`4. The ability to rehearse information depends on what occurs in the period between
`presentation and recall. For example, being asked to count backwards by 7s from
`119 is likely to block all rehearsal of a previously presented phone number.
`Rehearsal is an active process, so not rehearsing can also cause information loss.
`There is no statistical evidence on the extent to which real driving interferes with
`rehearsal of auditory information and how that should be accounted for in
`calculations, though such interference is believed to be minimal.
`
`5. When the capacity of the short-term store is exceeded, the information that is least
`important and/or oldest is deleted, depending upon user needs. For traffic
`information, the main road and the nearest exit or crossroad are priority items. In
`contrast, the number of cars in a crash is lower priority and more likely to be deleted
`when overload occurs.
`
`As a consequence of these rules, when people are asked to memorize a list of
`directions, the curve shown in Figure 2 often results. In brief, the first few items in the
`list are remembered because there are fewer items in the list when they are first
`encountered , so they can be rehearsed a greater number of times. For example, for
`the first item, that item is the only one to repeat. For the last few items, the time
`
`5
`
`FORD EX. 1022
`
`

`

`between presentation and recall is less than for items earlier in the list, so less decay
`occurs. One would expect to see such recall functions for auditory traffic messages.
`
`Recall
`Probability
`
`Primacy
`
`0
`
`I
`
`I
`
`I
`
`I
`
`I
`
`I
`
`|
`
`1st
`
`2nd
`
`I
`Last
`
`Item Number in List
`
`Figure 2. Recall probability for items in a list.
`
`Research Issues Investigated
`
`In selecting issues to examine, the following were considered:
`
`1. What issues were not addressed in the literature?
`2. Did the issue pertain to multiple systems?
`3. Would the results have applicability beyond the DIRECT project?
`4. Would the results have a practical impact?
`5. Would the results aid in the selection of a particular implementation?
`6. Was the issue reasonably inexpensive to evaluate?
`7. Could the issue be included in a clean experimental design?
`
`The literature contained considerable information on message recall. However, the
`effects of message quality have not been considered nor have the effects of listening
`to messages on driving performance. These issues were common to all of the systems
`of interest and, potentially, had major impacts on safety, usability, and usefulness of
`auditory traffic information systems. Accordingly, those topics were the focus of this
`research. Specifically:
`
`1. How does the recall of real messages vary
`
`(a) with the total amount of relevant information presented (number of terms)
`(b) if one is relevant
`(c & d) with the total amount of information presented: number of messages (c) and
`the number of terms (d)
`(e) with message quality
`(f) with driver differences (age and sex)?
`
`6
`
`FORD EX. 1022
`
`

`

`Of these issues, message quality has never been addressed in the literature.
`
`2. How does driving performance (mean speed, speed variance, mean headway,
`headway variance, lateral position variance) and driver control inputs (throttle
`variance, ste