Introduction - Experiment 1
Experiment 1 was performed to resolve two confounds in one of our road crossing situations – cars travelling from both left and right directions. Typically, when cars travel from two directions they travel along two lanes, however, when they travel from one direction, they typically travel from one lane. Therefore, any effects that might be associated with cars travelling from two directions instead of one cannot be separated. Moreover, cars travelling from both directions are travelling along both the near and the far lane, while cars travelling from one direction are travelling along the near or the far lane. Therefore, any effects of cars travelling along the far lane [5, 24, 25] cannot be separated from cars travelling along both lanes at once [24]. To resolve this, we examined crossing behaviour in OAs and YAs when cars travelled along one lane compared to two lanes, as well as cars travelling in the near lane, far lane, and two lanes.
In this experiment we also manipulated the speed of the cars and the view of the cars to determine whether these conditions would still have an impact on crossing decisions even when the crossing task was relatively simple (in comparison to Experiment 2).
Results - Executive function tests
Bootstrap t-tests and Bayes factors indicated that OAs and YAs had similar walking speeds (Table 1, Figure 1H), accuracy in the RMA task (Table 1, Figure 1A), local and global switch costs on RMA task accuracy (Table 1, Figure 1C, and E respectively), and BADS zoo map scores (Table 1, Figure 1G). Older adults showed significantly longer response times on the RMA task than YAs (Table 1, Figure 1B), as well as larger local and global switch costs on the RMA task than YAs (Table 1, Figure 1D, and F respectively).
Table1. Means, bootstrap t-tests, and Bayes Factors for the differences between OAs and YAs for the different EF measures, testing attention switching ability (local and global switch costs and RMA response times), spatial planning (BADS zoo map test), and walking speed. Significant results are highlighted.
|
Means
|
t-vale
|
df
|
CIs
|
p-value
|
d
|
Bayes Factor
|
Walking speed
|
YA: 1.33;
OA: 1.37
|
0.12
|
17.51
|
[-0.10, 0.12]
|
0.903
|
0.08
|
0.37
|
BADS
|
YA: 3.10; OA: 2.63
|
-1.68
|
12.25
|
[-1.77, 0.16]
|
0.099
|
0.42
|
1.63
|
RMA RT
|
YA: 1.42; OA: 1.93
|
3.05
|
17.36
|
[0.17; 0.85]
|
0.005
|
0.64
|
27.29
|
Local switch cost
|
YA: 0.23; OA: 0.38
|
2.68
|
10.79
|
[0.03; 0.34]
|
0.022
|
0.56
|
2.25
|
Global switch cost
|
YA: 0.40; OA: 0.94
|
2.98
|
14.39
|
[0.14; 0.74]
|
0.008
|
0.70
|
18.17
|
Results Experiment 1
Time To Impact (TTI)
There were no significant effects of cars coming from two lanes compared to one
lane; or cars travelling in the near versus far lane (Table 2, and Supplementary Table S1).
There were main effects of car speed and car view on TTI (Speed: β=-1.17, SE=0.53, t=-2.19, p=0.02; View: β=-1.85, SE=0.52, t=-3.53, p<0.001; Table 2, Figure 2D, and G). All participants had shorter TTI when cars travelled faster compared to slower (Figure 2G, Table 2). This reduction in TTI was larger for OAs, and participants with lower BADS zoo map scores (Figure 2H and I respectively; Supplementary Table S2). All participants decreased their TTI when cars were obscured compared to clear (Figure 2D, Table 2). This decrease was greater for OAs than YAs, and greater for participants with lower BADS scores compared to higher BADS scores (Figure 2E, and F; Supplementary Table S2).
There were main effects of age, BADS scores, and local switch cost on TTI (Table 2, Figure 2A, B, and C). OAs left more TTI than YAs (Figure 2A). Participants with high BADS scores, and participants with larger switch costs left less TTI than participants with low BADS scores or smaller switch costs (Table 2, Figure 2B).
There was an interaction between RMA RT and car speed on TTI (Table 2, Figure 2J). Participants with longer RTs on the RMA reduced their TTI by more than participants with shorter RTs on the RMA (Figure 2J).
Head movements
There were no significant effects of the lane the cars travelled (near vs far) or the number of lanes cars travelled in (one vs two) on the head movements made by participants (Figure 3, Table 2, Supplementary Table S3).
There was a main effect of age on sum head angle change with OAs turning their heads more than YAs (β = 110.13, SE = 43.04, t = 2.56, p = 0.014, Table 2, Figure 3A).
There was an interaction between local switch cost and view of the cars on the sum head angle changes made (β = 96.41, SE = 22.36, t = 4.31, p < 0.001, Table 2, Figure 3B). Participants with small local switch costs increased their head movements when cars travelled from an obscured view, while participants with large local switch costs decreased their head movements when cars travelled from an obscured view. Neither of these changes were significant (small local switch costs: β=1.71, SE=3.99, t=0.45, p=0.880; large local switch costs: β=-11.62, SE=7.21, t=-1.61, p=0.203; Figure 3B; Supplementary Table S4).
Table 2. LMM results for the TTI and the sum head angle change made by participants in Experiment 1. Significant results are highlighted in blue. Only significant main effects and interaction effects for car speed, view of the cars, age, BADS, local switch costs, and RMA RT are listed. Number of lanes and lane type (1 or 2) factors are listed as these are the main factors of interest for Experiment 1. For full LMM results see tables 1 and 3 in the Supplementary Materials. See Methods section for the models that were run.
Test
|
Factor
|
𝛽
|
SE
|
t-value
|
p-value
|
LMM: TTI
|
Lane number (1 or 2)
|
-0.43
|
0.64
|
-0.67
|
0.501
|
|
Near/Far lane
|
0.01
|
0.63
|
0.02
|
0.986
|
|
Car speed
|
-1.17
|
0.53
|
-2.19
|
0.029
|
|
Car speed * Age
|
-0.82
|
0.22
|
-3.68
|
<0.001
|
|
Car speed * BADS
|
0.30
|
0.09
|
3.25
|
0.001
|
|
View
|
-1.85
|
0.52
|
-3.53
|
<0.001
|
|
View * Age
|
0.52
|
0.22
|
2.37
|
0.018
|
|
View * BADS
|
-0.25
|
0.09
|
-2.81
|
0.005
|
|
Local switch cost
|
-1.76
|
0.85
|
-2.06
|
0.042
|
|
BADS
|
-0.47
|
0.18
|
-2.66
|
0.010
|
|
Age
|
1.02
|
0.42
|
2.43
|
0.018
|
|
RMA RT * Car speed
|
-0.68
|
0.30
|
-2.26
|
0.024
|
LMM: Head movements
|
Lane number
|
24.49
|
36.02
|
0.68
|
0.497
|
|
Near/Far lane
|
-28.47
|
35.92
|
-0.79
|
0.428
|
|
Age
|
110.13
|
43.04
|
2.56
|
0.014
|
|
Local switch cost * View
|
96.41
|
22.36
|
4.31
|
<0.001
|
Discussion Experiment 1
In Experiment 1 we manipulated the lane the cars travelled along (near/far), the number of lanes the cars travelled along, the speed of the cars, and the view of the cars (clear/obscure). Experiment 1 allowed us to resolve two potential confounds between the number of lanes and the number of traffic directions in Experiment 2.
We found no impact of cars travelling on the near vs the far lane or cars travelling along both lanes vs one lane on the crossing decisions or head movements made by participants. Therefore, any effects of the number of directions cars travel from (both vs one) in Experiment 2 will be due to the travel direction and not the number of lanes, resolving this confound.
In line with our previous findings [6], we found that OAs left more TTI than YAs, despite having the same average walking speed as YAs, suggesting OAs are able to make safe crossing decisions. We hypothesised that OAs left more TTI to compensate for reduced mobility such as slower walking times than YAs. However, the OAs in the current study did not have slower walking speeds than YAs so it would appear they do not need to compensate for this. OAs may leave larger TTIs than YAs to compensate for other factors such as being slower to initiate or prepare motor movements than YAs [5, 26, 27].
Overall participants with reduced EFs (both spatial planning and attention switching) left more TTI than participants with better EFs. This suggests that despite deficits in EFs participants are able to make safe crossing decisions, at least in simple situations.
When participants had less time to make a crossing decision (fast cars or obscured view), they left less TTI compared to when they had more time. This reduction in TTI was greater for OAs than YAs suggesting that OAs make less safe decisions than YAs when they have less time to make a decision. This may be due to OAs having slower visual processing speeds than YAs [10, 11, 12, 13]. Therefore, they may not be able to sample or process all the needed information to make a safe crossing decision in time and so they made riskier decisions.
We found that participants with poorer spatial planning abilities reduced their TTI by more than participants with greater spatial planning abilities when they had less time to make a crossing decision (fast cars or obscured cars). This suggests that participants with poorer spatial planning abilities made less safe decisions than participants with better spatial planning abilities when they had less time to make a crossing decision. This may be due to participants with poorer spatial planning abilities being less efficient at executing a plan than participants with better spatial planning abilities [28, 29].
Overall OAs moved their heads more than YAs. This may be due to OAs having reduced WM capacity [7, 8, 9] and therefore needing to turn their heads more to sample visually and refresh their knowledge of the vehicles’ trajectory.
We found that participants with poorer attention switching abilities moved their heads more than participants with better attention switching abilities when the view of the cars was obscured. A component of attention switching is disengagement [30]. If participants with poorer attention switching abilities are not able to disengage as quickly from the cars, they may follow them down the road for longer than participants with better attention switching abilities. This would lead to larger (greater angle) head movements being made.
Summary Experiment 1
We found that OAs were able to make safe crossing decisions as they leave more TTI than YAs. In specific situations, such as when participants had little time to make a crossing decision, OAs and participants with poorer spatial planning abilities begin to have difficulties in making safe crossing decisions as they left less TTI thus making less safe crossing decisions.
Introduction Experiment 2
In Experiment 1 we found that in situations where high processing speeds were required OAs and participants with poorer spatial planning abilities had difficulties making safe crossing decisions. Other situations that might lead to difficulties in making safe crossing decisions include high WM load situations such as when cars travel from multiple directions, when traffic density is high, or distractors are present. In Experiment 2 we examined the impact of these factors, as well as car speed, and cars travelling from an obscured view to determine if high WM load or short processing time conditions impacted OAs abilities to make safe crossing decisions.
Results Experiment 2
Table 3. LMM results for the TTI and the head angle change made by participants in Experiment 2. Only significant results are listed. For full LMM results see tables 5 and 7 in the Supplementary Materials. See the statistical analysis subsection of the Methods for the models that were run.
Test
|
Factor
|
β
|
SE
|
t-value
|
p-value
|
LMM: TTI
|
Car speed
|
-2.32
|
0.36
|
-6.92
|
<0.001
|
|
Car speed * Age
|
-0.78
|
0.15
|
-5.04
|
<0.001
|
|
Car speed * BADS
|
0.26
|
0.06
|
4.36
|
<0.001
|
|
Both directions * Age
|
-0.55
|
0.17
|
-3.33
|
<0.001
|
|
Both directions * BADS
|
0.15
|
0.06
|
2.43
|
0.015
|
|
Both directions * Local switch costs
|
1.85
|
0.37
|
4.98
|
<0.001
|
|
Car view
|
-1.70
|
0.39
|
-4.32
|
<0.001
|
|
Car view * Age
|
-0.72
|
0.17
|
-4.25
|
<0.001
|
|
Car view * Local switch costs
|
0.96
|
0.47
|
2.03
|
0.042
|
|
BADS
|
-0.54
|
0.16
|
-3.29
|
0.002
|
|
Age
|
2.04
|
0.39
|
5.27
|
<0.001
|
LMM: Head movements
|
Both directions
|
-72.46
|
21.06
|
-3.44
|
0.001
|
|
Both directions * Age
|
216.70
|
8.30
|
26.11
|
<0.001
|
|
Both directions * Global switch cost
|
25.90
|
9.94
|
2.61
|
0.009
|
|
Both directions * Local switch cost
|
69.07
|
15.74
|
4.39
|
<0.001
|
|
Both directions * BADS
|
10.16
|
3.47
|
2.93
|
0.003
|
|
Car speed * Age
|
33.32
|
6.76
|
4.93
|
<0.001
|
|
Car speed * Local switch cost
|
-43.78
|
12.88
|
-3.40
|
0.001
|
|
Car speed * BADS
|
-6.84
|
2.84
|
-2.41
|
0.016
|
|
View * Age
|
-20.78
|
8.32
|
-2.50
|
0.013
|
|
View * Local switch cost
|
11.21
|
3.96
|
2.83
|
0.004
|
|
View * BADS
|
-9.69
|
3.45
|
-2.81
|
0.005
|
|
Traffic density * Local switch cost
|
-94.48
|
15.70
|
-6.02
|
<0.001
|
Time to Impact (TTI)
There was a main effect of car speed on TTI (β=-2.32, SE=0.36, t=-6.92, p=0.000, Table 3, Figure 4A). All participants had shorter TTI when cars travelled quickly compared to slowly. The decrease in TTI was larger for OAs than YAs, and for participants with lower BADS zoo map scores than participants with higher BADS zoo map scores (Figure 4B, and C, Supplementary Table S6).
The LMM on TTI showed an interaction between age group and car travel direction, between spatial planning ability and car travel direction, and between attention switching ability and car travel direction (age group: β=-0.55, SE=0.17, t=-3.33, p=0.0001; BADS: β=0.15, SE=0.06, t=2.43, p=0.015; attention switching: β=1.85, SE=0.37, t=4.98, p=0.000, Table 3, Figure 4D, E, and F). All participants had longer TTI when cars travelled in both directions compared to just one direction. The differences were greater for YAs than OAs, for participants with higher BADS zoo map scores than participants with lower scores, and for participants with larger local switch costs than participants with smaller local switch costs on the RMA task (Figures 4D, E, and F, Supplementary Table S6).
The LMMs showed a main effect of view on TTI (β=-1.70, SE=0.39, t=-4.32, p=0.000, Table 3, Figure 4G). All participants decreased their TTI when cars travelled from an obscured view compared to a clear view. The decrease in TTI was greater for OAs than YAs and for participants with larger local switch costs than participants with smaller local switch costs on the RMA task (Figures 4H, and I, Supplementary Table S6).
The LMM also showed a main effect of age and BADS on TTI (Age: β=2.04, SE=0.39, t=5.29, p<0.001; BADS: β=-0.54, SE=0.16, t=-3.29, p=0.002. Table 3, Figure 4J and K). With OAs and participants with low BADS scores leaving more TTI than YAs and participants with high BADS scores (Supplementary Table S6).
Head movements
The LMM on head movement showed a main effect of car travel direction (β=-72.46, SE=21.06, t=-3.44, p=0.001; Table3, Figure 5A). All participants made more head angle changes when cars travelled from both directions compared to one direction (Figure 5). The increase in head movements were greater for OAs than YAs, participants with low BADS scores compared to high BADS scores, and participants with larger switch costs (local and global) compared to participants with lower switch costs (Figure 5B, C, D, and E, Supplementary Table S8).
The LMM on head movements showed a significant interaction between age group and travel direction, spatial planning ability and travel direction, and between attention switching ability and travel direction (Table 3, Figure 5J, K, and L). Participants with low BADS zoo map scores made more head angle changes when cars travelled from an obscured view compared to a clear one (Figure 5K; Supplementary Table S8). Participants with high BADS zoo map scores made fewer head angle changes when cars travelled from an obscured view compared to a clear one (Figure 5K, Supplementary Table S8). The change in head movements was not significant for OAs or YAs, or participants with smaller or larger local switch costs on the RMA task (Figure 5J and L; Supplementary Table S8).
The LMMs on the amount of head movements participants made showed interactions between car speed and age group, car speed and spatial planning ability, and car speed and attention switching ability (Table 3, Figure 5G, H, and I). OAs made more head angle changes, while YAs made fewer head angle changes when cars travelled quickly compared to slowly (Figure 5G, Supplementary Table S8). Participants with low BADS zoo map scores made more head angle changes when cars travelled quickly compared to slowly (Figure 5H, Supplementary Table S8). Participants with high BADS zoo map scores did not significantly change their head movements (Figure 5H, Supplementary Table S8). The LMM on the sum head angle change participants made also showed a significant interaction between car speed and local switch costs on the RMA task but the simple effects LMMs showed that this was not significant at each level of the local switch costs measure (Figure 5I, Supplementary Table S8).
The LMM on head movements showed an interaction between traffic density and local switch costs on the RMA task (Table 3, Figure 5F). Participants with small local switch costs made less head angle changes when traffic density was high compared to when it was low (Figure 5F, Supplementary Table S8). Participants with large local switch costs did not significantly differ in the head movements they made when traffic density was high compared to when traffic density was low (Figure 5F, Supplementary Table S8)
Discussion Experiment 2
In Experiment 2 we examined how the complexity of road crossing situations impacted crossing decisions made by OAs and YAs with varying levels of EFs. We manipulated the complexity of the road crossing situation by varying car speed, the direction the cars travelled in (both or one direction), the initial viewpoint of the cars (clear vs obscure), the traffic density, and the presence of pedestrian distractors.
In a replication of Experiment 1 we found that overall OAs left more TTI than YAs, and that participants with poorer spatial planning abilities left more TTI than participants with better EFs. We also replicated the finding that all participants reduced their TTI when they had less time to make a crossing decision than when they had more time (fast vs slow cars and obscure vs clear view), which was amplified for OAs and participants with poorer spatial planning abilities. We discussed these results in the discussion section of Experiment 1.
When cars travelled from an obscured view, participants with poorer attention switching abilities reduced their TTI by more than participants with better attention switching abilities. Attention switching involves the process of disengaging with a previously attended task or stimuli, updating WM with a new task or stimuli, and inhibition of distracting stimuli [31, 32]. Previous studies have shown that updating WM after a delay leads to poorer performance on the attention switching task than with no delay [33]. In the case of an obscured view there was a delay in seeing the car compared to when the view was clear. Therefore, participants with poorer attention switching abilities may make less safe crossing decisions in the obscured view condition because they take longer to update their WM with new information about car position giving them less time to make a crossing decision.
Participants with poorer spatial planning abilities increased their head movements when cars travelled from an obscured view than from a clear view. The opposite was seen for participants with better spatial planning abilities. These head movement changes were not associated with a change in crossing decisions. Spatial planning involves anticipating where objects will appear along a route [34]. To make a safe crossing decision when the view of the cars was obscured participants needed to anticipate where the cars would become visible along the road and approximate the speed of the cars. Participants with better spatial planning abilities may be better at anticipating where the cars will become visible and so they focus their attention on this location of the road. However, participants with poorer spatial planning abilities may not be able to anticipate the location the cars will become visible as easily so they switch their attention between multiple locations where they think the car could become visible, leading to more head movements.
Some road crossing situations were handled well by all participants. When cars travelled from both directions at once all participants increased their TTI. This suggests that participants were able to identify that this was a more dangerous situation and they approached the road crossing more cautiously. The increase in TTI was greater for YAs than OAs. This could be due to YAs faster visual processing speed and so they were able to take in all the information faster than OAs, allowing them to leave more TTI than OAs. This suggestion is supported by the finding that OAs made more head movements when cars travelled from both directions than when cars travelled from one direction. If OAs had to make multiple glances in the two car directions, compared to YAs only needing one glance in each direction, then YAs need less time to recognise the situation as more dangerous, allowing them to leave more TTI than OAs.
Making more head movements may not be an entirely bad strategy. Participants with better spatial planning abilities and participants with poorer attention switching abilities made more head movements than participants with poorer spatial planning abilities and participants with better attention switching abilities when cars travelled from both directions. These increased head movements were associated with an increase in TTI. Therefore, it would seem that these groups of participants have come up with an attentional strategy that involves a lot of head movements but allows them to sample enough information about the crossing situation to make safe crossing decisions. More research is needed to elucidate the different strategies used between the groups, and how these strategies may aid in making safe crossing decisions for one group (better spatial planning / poorer attention switching) but not another (OAs).