Abstract: Change blindness typically refers to the phenomenon where observers fail to realize and detect changes in objects or scenes when there is a brief visual interruption, restriction, or disturbance in the visual environment. This phenomenon is considered a significant contributor to driver human factors errors, posing a serious threat to public safety. Prior research has often relied on static masking paradigms, which may not adequately reflect change detection during natural, full-field observation. Furthermore, by focusing on variables tied to a single cognitive mechanism, prior studies have struggled to examine the roles of attention and memory representations within a unified framework in complex traffic scenarios. This study aimed to examine how attention and memory representations affect change blindness in response to sudden shifts in the motion path of traffic-scene images.
The method of global motion direction transients was adopted to induce change blindness, employing images of real traffic scenes with different change types as experimental stimuli. The image moved along a certain path and changed instantaneously either at the moment of turning or at the midpoint of the straight line. In Experiment 1, a within-participants design was conducted with 2 (movement speed: fast, slow) × 2 (movement path: turning motion, straight motion) × 3 (change type: deletion, addition, location change). Forty participants were asked to detect the target changes. The results showed that fast motion significantly exacerbated change blindness solely under turning conditions, but had no significant effect during straight motion. In Experiment 2, the fast movement speed from Experiment 1 was used, a within-participants design was adopted, featuring 2 (movement time: long, short) × 2 (movement path: turning motion, straight motion) × 3 (change type: deletion, addition, location change). Thirty participants were requested to detect the target changes. The results indicated that longer encoding time for memory representations could mitigate change blindness within a limited scope. In Experiment 3, the fast movement speed was also used, a within-participants design with 3 (movement path: high-expectation turning motion, low-expectation turning motion, straight motion) × 3 (change type: deletion, addition, location change) was employed. Thirty-six participants were required to detect the target changes. The results suggested that individuals exhibited better detection performance for target changes under high-expectation conditions.
The main findings are as follows. (1) Increased movement speed exacerbated change blindness, likely because it intensified competition for attentional resources when the direction of movement suddenly changed. However, movement speed needed to interact with other basic visual feature changes to induce change blindness. (2) Prolonging movement time, which refers to increasing encoding duration, could alleviate change blindness. Visual processing time might enhance change detection by improving the precision of scene memory representations during encoding; however, this effect appeared to be limited. (3) Heightened expectancy levels reduced change blindness. Individuals’ spatial anticipation of the turning direction likely introduced a response bias, endogenously modulating the pre-allocation of attention and the effective encoding of memory representations, thereby improving the quality and efficiency of change detection.
In summary, change blindness in traffic-scene images during global motion direction transients arises from the interaction between competition for attentional resources and the limited capacity of memory representations. Expectations may endogenously modulate attentional allocation and the effective encoding of these representations, thereby influencing the likelihood of change blindness. These findings provide empirical support for and extend the attention-representation account of change blindness.

Key words: change blindness, visual search, motion-induced blindness (MIB), attention, memory representation