Individual Differences in Intramodal and Crossmodal Inattentional Insensitivity and the Design of In-vehicle Alert Systems

Bridget A Lewis

Advisor: Carryl Baldwin, PhD, CHSSWeb Design Preview

Committee Members: Tyler Shaw, Eva Weise

David J. King Hall, #2072
December 01, 2017, 03:00 PM to 05:00 PM

Abstract:

Around 20% of all automobile crashes in recent years have been linked to driver distraction or inattention.  A subset of these crashes, involve “Looked But Failed to See” (LBFTS) incidents in which an otherwise attentive driver completely fails to notice a salient signal.  In the best case, this may involve a driver putting on the brakes late because she failed to notice a red light and stopping with her nose into an intersection, causing embarrassment but no harm.  But in the worst case, looking but failing to see causes about 6% of all injury and fatality-related crashes per data from the Fatal Accident Reporting System (FARS) maintained by the National Highway Traffic Safety Administration (NHTSA).  LBTFS are a type of inattentional blindness, a subset of inattentional insensitivity.  Inattentional insensitivity is a blanket term that describes the well-known phenomena of inattentional blindness and inattentional deafness.  These phenomena occur when an otherwise salient stimulus is missed during high levels of perceptual load.  For example, pilots coming in for a difficult landing in high cross winds miss auditory alarms notifying them of a landing gear failure, or drivers lost at night on an unfamiliar route fail to see a stop sign.  It is hypothesized that inattentional insensitivity is integrally tied to an individual’s working memory capacity.  Previous studies have proposed theoretical accounts involving both single and dual routes for this relationship.  A major goal of this dissertation is to determine which of these theoretical explanations best predict patterns of inattentional insensitivity.  A second goal of this dissertation is to address methods of ameliorating inattentional insensitivity, regardless of their cause, via the design of effective, multimodal alert systems.  

Towards the second goal, the first study, details an investigation of multimodal urgency scaling with the goal of determining perceived changes in urgency relative to physical changes in visual, auditory and tactile stimuli.  Psychometric functions were obtained for various parameters within each modality based on perceptions of urgency, annoyance and acceptability.  Results indicated that auditory stimuli affected the biggest increases in urgency relative to physical changes, but that with increased urgency often came increased annoyance.  Visual stimuli were rarely rated as annoying but were also unable to achieve similarly high levels of urgency relative to auditory or tactile stimuli.  Tactile stimuli showed the greatest utility (indicating greater urgency changes in relation to annoyance changes).  

The second study, was designed to validate the psychometric functions established in the first study by examining behavioral responses to warnings designed to be perceived as highly urgent and time critical versus warnings missing key parameters within the context of driving.  Specifically, the second study examined the potential for appropriate warnings to eliminate inattentional insensitivity to alerts while distracted, regardless of why it was occurring.  Towards this aim, the second study required participants to drive a simulated course while completing a distracting task and following a lead vehicle.  At a pre-set point, the lead vehicle swerved sharply into the left lane to avoid a revealed, stopped car.  Participants then received either a “good” warning, one that met all pre-defined criteria, an “edge” warning, one that met only some of the criteria, or no warning.  Results indicated that, while crash occurrences were not significantly different, for those who did crash, they crashed at a significantly slower speed.

The final study in this series, sought to examine the effect of working memory capacity (WMC: as measured by OSpan) on various types of inattentional insensitivity.  Specifically, inattentional insensitivity was examined for intramodal and crossmodal tasks involving either visual or auditory critical signals.  Participants were asked to first complete a computerized version of the OSpan task to evaluate their working memory capacity.  They were then assessed for inattentional insensitivity starting in one of four conditions: visual task- visual critical signal, visual task-auditory critical signal, auditory task-auditory critical signal, and auditory task-visual critical signal.  In this series, the visual task was a cross arm-length detection task and the auditory task was a rapid serial auditory presentation task.  The critical signals (CSs) were either visual pictures of shapes or auditory names of shapes.  Results from Study 3 indicate that, although inattentional insensitivity was present in all modality combinations for some proportion of the population, individuals were less likely to miss critical signals when they were in the same modality as the main task (intramodal signals).  Results also indicate a significant difference in sensitivity by WMC, where those with medium to high WMC were significantly more likely to notice an intramodal CS than a crossmodal CS, though this effect was not present for those with low WMC levels.

Results from the present series of studies inform the design of in-vehicle alerting systems, and may be particularly pertinent for highly automated or autonomous vehicles.  The drivers or, for lack of a better word, operators of these vehicles may well be fatigued, distracted, or otherwise impaired which will increase the need for targeted, highly effective warnings when operator intervention is required.