It's All in the Timing: Using the Attend Algorithm to Assess Texting in the Naturalistic Driving Study
DOI: 10.17077/drivingassessment.1665
archive: archived pipeline: cataloged verified
Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)
Summary
This study investigates the temporal dynamics of driver distraction, specifically examining how cellular phone texting and radio interactions relate to safety-critical events (SCEs) in naturalistic driving. The authors address the limitation of aggregating rich naturalistic driving data into single data points by utilizing the AttenD algorithm, which models situation awareness as a buffer that depletes during off-road glances and replenishes during on-road glances. The research aims to determine whether crash-related distraction differs from baseline distraction in terms of duration, glance patterns, or the timing of task initiation relative to a precipitating event. The analysis used the Naturalistic Engagement in Secondary Task (NEST) database, a subset of the SHRP2 dataset containing 20-second epochs of glance and secondary-task data. The study compared Crash epochs (ending in a crash or near-crash) with Baseline epochs (no crash) for two secondary tasks: texting and radio adjustment. The AttenD algorithm was applied to sample-level glance data (10 Hz), calculating a buffer value that decreased by 1 point per second of off-road glancing and increased by 1 point per second of on-road glancing, capped at 2 points. The final dataset included 22 Crash and 69 Baseline texting epochs, and 13 Crash and 59 Baseline radio epochs, with drivers averaged to ensure independence. Results indicated that texting duration and task frequency were statistically similar between Crash and Baseline epochs. However, the temporal alignment differed significantly: Crash texting tasks ended closer to the onset of the precipitating event. Crucially, the AttenD buffer analysis revealed that Crash texting epochs began with a significantly lower initial buffer value (mean 1.45) compared to Baseline texting (mean 1.70), indicating that drivers initiated texting with already depleted situation awareness. In contrast, radio interactions showed substantial differences in duration and buffer depletion between Crash and Baseline conditions, suggesting different task natures rather than just timing issues. Crash radio interactions lasted longer and resulted in lower final buffer values. The authors conclude that the risk of texting-related crashes is not primarily driven by the duration of the task or the total amount of glancing, but by the timing of the task initiation relative to the driver’s current situation awareness. Drivers who initiate texting when their AttenD buffer is already low are more likely to engage in ill-timed glances that coincide with rapidly changing road conditions. This suggests that crash causality involves a cascading loss of situation awareness, where prior secondary tasks or poor judgment in task initiation compromise the driver’s ability to monitor the road effectively. The study highlights the utility of the AttenD algorithm in identifying these temporal patterns and suggests future work should explore individual differences and comorbid tasks contributing to low initial situation awareness.
Key finding
Texting periods that ended in crashes were initiated at a lower level of situation awareness and occurred closer to the precipitating event compared to baseline texting, suggesting that timing and pre-existing cognitive load are critical determinants of crash risk.
Methodology
naturalistic
Sample size: 165
Provenance
The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed. Discovered via openalex_abstract on 2026-05-08.
| Stage | Outcome | Tool | Model | Prompt | Attempts | Completed |
|---|---|---|---|---|---|---|
| discover | success | — | — | — | 1 | 2026-05-07 |
| archive | success | canonical_url | — | — | 1 | 2026-06-06 |
| extract | success | cached | — | — | 3 | 2026-06-10 |
| clean | success | clean | — | — | 1 | 2026-06-04 |
| chunk | success | chunk | — | — | 1 | 2026-06-04 |
| embed | success | embed | Qwen/Qwen3-Embedding-8B | — | 1 | 2026-06-04 |
| enrich | success | openalex | — | — | 2 | 2026-05-08 |
| promote | success | — | — | — | 1 | 2026-05-07 |
| summarize | success | llm | qwen3.6-27b-prismaquant | summ-v5 | 2 | 2026-06-10 |
| tag | success | vector_similarity | — | — | 15 | 2026-06-11 |
| verify | success | — | — | — | 2 | 2026-06-10 |
Summary generated by qwen3.6-27b-prismaquant on 2026-06-10; verification: verified.
Topics
Ranked by relevance to this paper. Hover a topic for its definition.
Information type
What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).
- Empirical Findings: observational prevalence, behavioral performance data, crash risk outcomes