A method to measure Representativity and Univocity of traffic signs and to test their effect on movement.

OBJECTIVE: The present protocol aims to understand how participants represent traffic signs (Psychology of Thinking) and how the attention of these signs-and the way of processing them-can influence on their path movement (Motor Processes). Knowing how humans process the meaning of signs (not just by learning but instinctively understood) will improve reaction times and decision making when driving.
BACKGROUND: In laboratory tasks, a number of models have attempted to explain the general relationship between attention and movement. The cornerstone of the effects is found on the meaning of attentional cues.
METHOD: By using a tracking task, the influence of traffic signs on movement is been tested.
RESULTS: Results point out that the signs least representative of their meaning produce a greater deviation from the center of the simulated road than the most representative signs.
CONCLUSIONS: The economic, social and psychological consequences of car accidents are well-established. Every single effort orientated to amend this social problem is welcome. Taking into account the results reported in this work, it is recommended that the traffic signs are designed as much representative of their meaning as possible. APPLICATION: The methodology used in this study can be applied to testing the Cognitive Ergonomics of signposting on roads; analyzing, classifying, and discarding the traffic signs that produce counterproductive effects on movement from the current Manuals of Traffic Signposting all around countries.•An original methodology has been generated for classifying traffic signs, which has not been never tested in the literature.•The well-consolidated tracking task is used for evaluating their effect on movement.•The main result is that traffic signs, that do not represent properly their meaning, provoke a significant and dangerous deviation from the drivers' path.

Specifications Table

Method details

The present method is a protocol in which three experiments, first of all, classify traffic signs and, afterward, check the possible effect of representativity and univocity of signs on movement. However, k-means analysis and ANOVA are not new.

Experiment 1 (to classify traffic signs)

In this experiment, Reaction Times (RTs) are measured in response to the presentation of a serial of traffic signs. Participants have to decide if a previous definition of a sign corresponded to the sign presented at that moment.

The representativeness of signs is defined as the degree to which they reflect features of their parent population. On the other hand, univocity is defined as the absence of polysemy (more than one meaning).

Therefore, these two concepts are operationalized as (cf. [[1], [2], [3]]): (a) if the sign really represents its meaning (representative sign), before the right matched definition-sign, RTs will be shorter than when the sign does not represent that meaning; (b) if the sign is not ambiguous (univocal sign), before the wrong matched definition-sign, RTs will be shorter than when the sign is ambiguous.

Stimuli

From the population of traffic sign icons specified in any “Manual of Traffic Signposting” of any National Road Safety Commission, informative, warning or regulatory signs can be selected for the study (e.g., Fig. 1). It can be used a selection criterion based on whether the element–that the sign tries to represent–is inserted either on the road or nearby. Distant elements from the road could be discarded. The reason for this decision is that mental models of the elements (e.g., [4]) closer to the route are more likely to cause an interference on movement [5]. Within this criteria, the secondary criterion can be: Same signs with different colors are randomly selected (since the color has not significant effect on the representativity or the univocity of traffic signs; [[1], [2], [3]]). The meaning of the sign is independent of the color they have (see Fig. 2). For example, white, green and blue colors just indicate the type of the road (regional road, freeway, or highway respectively) where the drivers are traveling on. On the other hand, the definition of signs can be extracted from their legal frame. Table 1 presents the translation of examples of definitions regarding informative signs.

Fig. 1

Informative sings tested in their representative and univocal meaning.

Fig. 2

Example of the meaning of the sign independently of the color printed in.

Note. The meaning of the sign is proximity of a public phone.

Table 1

Translation of the sign definitions used in the experiment 1.

Codification
Sign	Definition
I1	Route to follow in an intersection
I2	Coming emergency track
I3	Perpendicular route exclusive for public transport
I4	Direction to abandon the road
I5	Distance to the next way out
I6	Confirmation to the coming destinations and their distances
I7	Pan-American highway
I8	Number of the highway you are driving in
I9	Name of the street and point on the road
I10	Interest point announcement
I11	Restaurant
I12	Accommodation
I13	Cash machine
I14	Service station
I15	Restroom
I16	Public phone
I17	Freeway beginning
I18	Freeway ending
I19	Side way out
I20	Last way out before freeway
I21	U-turn
I22	Place to park
I23	Emergency phone
I24	Freeway with “telepayment”a
I25	Tollbooth
I26	Weigh station
I27	Bus stop
I28	Radar control

Note.

System to automatically pay the turnpike.

Materials and task

A plain computer can be used to present stimuli and record participants’ responses, by means of OpenSesame™ program [7].

In a computer screen, definitions of signs are presented after a fixation cross (see Fig. 3), this presentation lasted as long as the participant does not push the spacebar key. These definitions could match or not with the actual sign presented afterward. The definitions, that mismatch their signs, are presented randomly. Participants have to decide if the definition corresponded or not to the sign, by pressing “s” (congruence trials) or “k” keys (incongruence trials) respectively, as an example.

Fig. 3

Schematic of the decision task.

Note. Presentation sequence of the task: fixation point, match/mismatch definition, fixation point, sign to be evaluated.

Procedure

Once participants are seated comfortably (and after having fulfilled the informed consent form), they are given the instructions of the task. The instructions specify that, on a single trial, any given definition of a traffic sign will be presented and, after the sign disappearance, they have to decide if that sign really represented the meaning of the definition previously shown.

The signs are presented with their match definition or with any other mismatch definition (randomly extracted from a population of definitions). Therefore, for each sign, there are two kinds of trial (congruent/incongruent). A complete block is constituted by the match and the mismatch trials. The total number of blocks can be decided depending on the total number of traffic signs selected.

Data analysis

All RTs are submitted to a k-means analysis to cluster signs with equivalent responses. K-means is a vector quantization in which every observation belongs to the cluster with the nearest centroid [6]. This centroid is calculated based on the distribution of data, not the mere mean. This analysis make it possible to cluster not only signs (observations) that have similar means but signs that have similar pattern of response; similar distribution of TRs. In this sense, two kinds of RTs, regarding the concepts of representativity and univocity, are analyzed. On the one hand, TRs for the matched definition-sign are clustered on Representative signs and Non-representative signs. On the other hand, for the mismatched definition-sign, RTs are clustered on Univocal signs and Ambiguous sign. One-way ANOVA test is carried out to compare the differences between both created groups of signs in their RT means between the both groups within the both concepts (representativity and univocity).

In order to check the significant relationship between representative/non-representative signs and the number of errors committed in the decision task for each group, the contingent table for these matches is submitted to χ² test; ϕ coefficient is used to measure the magnitude of that correlation.

Experiment 2 & 3

Overview

Taking into account the different cluster groups created in Experiment 1, the most “representative” signs of each group (the ones closest to the center of both cluster within the both constructs of representativity and univocity; see Table 2, Table 3 and Fig. 4 as an example) are separately introduced into a tracking task [8]. Since, when individuals do not understand any piece of information, they move closer to the source of that information (e.g., [5]), the first hypothesis is that the non-representative signs produce a differential influence on movement, as shown in previous literature ([[1], [2], [3]]). This effect will take the form of a significant greater deviation (from the center of the road) for non-representative signs than for representative signs. The second hypothesis is that, since univocal and ambiguous signs have, at least, one meaning, these signs would not deviate significantly the movement to them. In Experiment 2 and 3, the paradigm of Vilchez [5] to measure the influence of the attention-to-stimuli on movement is used.

Table 2

K-means clustering based on the RTs to respond to each sign.

Representativity
Representative signs		Non-representative signs
Distance to cluster center	Sign	Distance to cluster center	Sign
357.38	I28a	632.44	I7a
404.60	I21a	715.56	I18a
407.32	I14a	730.59	I24a
462.97	I12a	835.46	I17a
476.82	I15a	884.72	I3a
495.03	I23a	912.46	I5a
501.75	I6a	966.64	I25a
523.70	I11a	1007.44	I22a
571.25	I13	1025.04	I1
619.94	I16	1040.55	I4
673.84	I8	1087.58	I2
675.63	I27	1141.75	I26
702.94	I19	1548.92	I20
750.62	I9
895.99	I10

Note.

Signs selected in each group (Representative signs and Non-representative signs) for experiment 2.

Table 3

K-means clustering based on the RTs to respond to each sign.

Univocity
Univocal signs		Ambiguos signs
Distance to cluster center	Sign	Distance to cluster center	Sign
238.67	I13a	521.976	I17a
265.72	I10a	634.039	I1a
266.72	I14a	922.080	I18a
306.32	I21a	937.046	I5a
321.71	I9a	2158.49	I3a
323.4	I20
339.44	I28
341.81	I7
343.82	I11
346.24	I15
363.64	I16
366.31	I23
368.32	I12
371.12	I25
404.7	I27
443.5	I6
470.21	I22
473.67	I2
623.52	I4
633.08	I24
650.15	I8
656.78	I19
901.55	I26

Note.

Signs selected in each group (Univocal signs and Ambiguous signs) for experiment 3.

Fig. 4

Selected signs in representativity (representative signs and non-representative signs) and univocity (univocal signs and ambiguous signs).

A 60-pixels-wide, gray road is presented on 800-pixels-wide, green backgrounds in a computer screen. The simulated road is limited by two continuous, white lines and divided in two by a 3-pixels-wide, discontinuous, white line (see Fig. 5). Two kinds of background were presented: non-experimental and experimental backgrounds. Non-experimental backgrounds are created with several routes where the road turned right and left (see [5], for a detailed description of the methodology).

Fig. 5

On the left, example of non-experimental background. On the right, example of experimental background.

The participants controlled a green circle of 22-pixels-wide in diameter, with a black circumference and a central dot (see Fig. 6). Traffic signs (see Fig. 4) were superimposed on both sides of the road before a Y-junction in experimental backgrounds (see Fig. 7 as an example).

Fig. 6

Car controlled by the subject (green circle).

Fig. 7

Example of each experimental conditions. From left to right and up to down: (a) left-far condition (upper-left); (b) left-close condition (upper-right); (c) right-close condition (lower-left); and (c) right-far condition (lower-right).

Materials and tracking task

A plain computer can be used to present stimuli and recorded participants` trajectory movement. Participants drive the green circle stimulus with a steering wheel. The steering wheel movements modified the position of the center of the green circle on the x-axis. Its position on the y-axis is fixed at a specific location (see [5], as a reference). Backgrounds ran across the screen, giving the feeling of movement.

The aim of the task is to track the center of the simulated road with the green circle during proximately 20 min. The road is designed by using a randomized sequence of backgrounds. However, the common denominator is the sequential appearance of non-experimental/experimental background. In every trial, only one sign is presented at the same time, in any side of the road, and only in experimental backgrounds. The sign appears approximately 123 pixels before the Y-junction, 70 pixels from the center of the road, and during 6 registered frames (each frame with an average duration of 60 ms).

Once participants are seated comfortably, and after having fulfilled the informed consent form, they are given the instructions of the task. The instructions specify the possible appearance of a distractor (the traffic signs) during the performance of the task and the free-option to take the route that participants wanted to in the Y-junction; in experimental backgrounds (see Fig. 5 as a reference).

The total number of data rows (or registered frames) is exactly 17,601 per participant. Data was filtrated by experimental conditions and different measure stages (previous or subsequent to the signs presentation). In the first measurement stage, the space right before the picture presentation (pre-cue stage or A frames) is codified as General-movement control condition. During the second measurement stage–the space where the sign is being presented (cue-appearance stage or B frames)–, the codification depends on the experimental conditions of that particular trial. In the third measurement stage–the space where the sign disappears (post-cue stage or C frames)–, the codification depends on the experimental condition of the previous second stage.

The error is calculated by subtracting, frame by frame, the position of the center of the road from the x-position of the green circle. Repeated measures ANOVA 2 × 2 (Position x Representativity) is conducted on every experimental frame belonging to cue-appearance (B frames) and post-cue (C frames) stages. Pre-cue General-movement control stage (A frames) is not included in the analysis because, in this stage, the different experimental conditions are not presented.

On the other hand, the two possible routes to take in the Y-junction in experimental backgrounds are also analyzed. To this end, the count of times every participant choses a specific option is weighted to the total number of possible choices. Data is subjected to a Kendall´s W coefficient of concordance test, focusing only on one option (right route) because the branch options are symmetrical, that is to say, taking the right route is equivalent to non-taking the left route within a specific trial.

Subject area	Select one of the following subject areas:• Engineering • Neuroscience • Psychology • Social Sciences
More specific subject area	Cognitive Ergonomics
Method name	A protocol to analyze the Representativity and Univocity of traffics signs
Name and reference of original method	If applicable, include full bibliographic details of the main reference(s) describing the original method from which the new method was derived.
Resource availability	If applicable, include links to resources necessary to reproduce the method (e.g. data, software, hardware, reagent)