Fixating on the size-speed illusion of approaching railway trains: What we can learn from our eye movements

•The size-speed illusion may be due to observer’s eye movements. Manipulation of eye movements has been effective in reducing this illusion.•We tested observers perceived speed of vehicles (trains and cars), while requiring them to fixate on a static object (square) in the foreground.•Placing the fixation square around the front region of the moving train reduced the size-speed illusion.•However the illusion persisted when the square was placed around the train’s visual centroid region.•Our findings are robust and we suggest that interventions based on this strategy should be designed and tested in future studies. Railway level crossing collisions have recently been linked to a size-speed illusion where larger objects such as trains appear to move slower than smaller objects such as cars. An explanation for this illusion has centred on observer eye movements – particularly in relation to the larger, longer train. A previous study (Clark et al., 2016) found participants tend to make initial fixations to locations around the visual centroid of a moving vehicle; however individual eye movement patterns tended to be either fixation-saccade-fixation type, or smooth pursuit. It is therefore unknown as to which type of eye movement contributes to the size-speed illusion. This study isolated fixation eye movements by requiring participants to view computer animated sequences in a laboratory setting, where a static fixation square was placed in the foreground at one of two locations on a train (front and centroid). Results showed that even with the square placed around the front location of a vehicle, participants still underestimated the speed of the train relative to the car and underestimation was greater when the square was placed around the visual centroid of the train. Our results verify that manipulation of eye movement behaviour can be effective in reducing the magnitude of the size-speed illusion and propose that interventions based on this manipulation should be designed and tested for effectiveness.