Evaluation of associative learning methods to train drivers to give way to motorcyclists

Monash University Accident Research Centre - Report #278 [2007]

Authors: Simon Hosking, Christine Mulvihill, Jessica Edquist, Michael Lenné, Karen Stephan, Mark Symmons, Cameron Murdoch and Jeff Archer

Full report in .pdf format [637KB]


A high proportion of motorcycle casualty crashes involve drivers failing to give way to on-coming motorcycle riders at intersections. Previous research has shown that associative learning interventions can be used to successfully change driving-related decisions by manipulating learned relationships between crash risk and motorcycles. For example, Harrison (2005) demonstrated that the effect of increasing the crash risk associated with approaching motorcycles in a PC-based task led to increased wait responses to photographs of motorcycles at intersections, and that these effects were maintained in a four-week follow-up test. The present experiment was a partial replication and extension of Harrison's study that aimed to (i) assess the associative learning approach in dynamic learning situations where the task was more representative of real driving, (ii) demonstrate transfer of learning from the static learning contexts to dynamic contexts that more closely simulate the judgements made when turning at intersections, and (iii) test the longevity of learning after a period of 12 weeks after a training phase. The results replicated Harrison's finding that associative learning interventions increase the likelihood of drivers to respond that they would wait for approaching motorcycles at intersections. However, such learned avoidance responses only occurred when the associative learning intervention used photographic stimuli, and the increase in learned wait responses that occurred for photographic stimuli in the initial phase of the experiment was extinguished in a follow-up test using video stimuli. When the intervention used video stimuli, there was no evidence of increased avoidance responses to motorcycles. Rather, the probability of a wait response to cars decreased in a manner that was consistent with inhibitory learning, and this latter effect persisted in a four-week follow up test. On the basis of these results, an additional 12 week evaluation was not conducted. Conclusions and recommendations about the associative learning technique, and other methodologies, for training drivers to give way to motorcycle riders at intersections are provided.

Sponsoring organisation - VicRoads