Trajectory changes are susceptible to change blindness manipulations

People routinely fail to notice that things have changed in a visual scene if they do not perceive the changes in the process of occurring, a phenomenon known as ‘change blindness’ (1,2). The majority of lab-based change blindness studies use static stimuli and require participants to identify simple changes such as alterations in stimulus orientation or scene composition. This study uses a ‘flicker’ paradigm adapted for dynamic stimuli which allowed for both simple orientation changes and more complex trajectory changes. Participants were required to identify a moving rectangle which underwent one of these changes against a background of moving rectangles which did not. The results demonstrated that participants’ ability to correctly identify the target deteriorated with the presence of a visual mask and a larger number of distractor objects, consistent with findings in previous change blindness work. The study provides evidence that the flicker paradigm can be used to induce change blindness with dynamic stimuli, and that changes to predictable trajectories are detected or missed in the similar way as orientation changes.


Introduction
People routinely fail to notice that objects have changed in a visual scene if they do not 27 perceive the changes in the process of occurring, a phenomenon known as 'change blindness' 28 (1,2). The majority of lab-based change blindness studies use static stimuli and require 29 participants to identify simple changes such as alterations in stimulus orientation or scene 30 composition (3-5), though others use more complex and realistic environments, especially 31 driving simulators (6-8). This study examines whether changes to dynamic properties are 32 detected or missed in the same way as changes to static properties.

33
Changes to static properties (e.g. the presence of a stimulus, or its orientation) are most 34 readily detected when the transients (moment-to-moment variations) accompanying a change 35 prompt an explicit comparison between a stored representation of a stimulus and its current 36 presentation (9-11). Change blindness frequently occurs when this process is disrupted. If the 37 representation of a stimulus includes information on dynamic properties (e.g. the trajectory 38 along which a stimulus is travelling), change blindness would be expected to occur for changes 39 to dynamic as well as static stimulus properties. Common methodologies for inducing change 40 blindness prevent transient registration, typically by masking (12) or eliminating transients 41 (1,13,14). In addition to masking transients, exhaustion of working memory capacity is 42 required to produce change blindness effects reliably (15), with the contents of working 43 memory exhibiting resistance to change blindness (2,16), a phenomenon which is stable 44 enough to allow change blindness task performance to act as a guide to working memory 45 contents in attentional bias studies (17-19). 46 Change blindness as deployed in the study of other phenomena (e.g. attentional biases) is reasonably well understood, but a gap exists between these structured laboratory experiments 48 and the more sophisticated simulator-based and natural-world experiments (1,20,21  115 The task page presented participants with a 700x700 pixel working area. The initial stimulus 116 consisted of a number of 50x25 pixel rectangles with randomly selected colours moved at 150 117 pixels/second along a straight-line trajectory (Fig 1a).  Research on multiple object tracking (28) has shown that, although object tracking does not are to some extent resistant to change blindness (31,32).;tracking target selection is typically 149 exogenous and based on colour and spatial location (33), As both of these were randomised 150 in all trials, there was no systematic relationship between the likelihood of an object being 151 tracked and its being the target object for that trial. Therefore, the dynamic paradigm did not 152 undermine the validity of the change blindness.

153
Participants were asked to press the spacebar as soon as they had identified the altered 154 rectangle. Pressing spacebar halted the movement of the rectangles and correct identification 155 was checked by requiring the participants to click the rectangle which had changed.

156
Participants were given the opportunity to practice the task until they were satisfied with their 157 performance, and were provided with feedback as to their accuracy during the practice.

158
Each trial had one of eight possible types, defined by its specific arrangement of three different 159 binary variables: whether a mask was present or not; whether scene load was low (2 160 rectangles) or high (6 rectangles); and whether the target rectangle was changed in orientation 161 or trajectory. Experimental condition was selected randomly at the beginning of each trial. The 172 Participants were invited to complete as many trials as they wished, though the provision of 173 full feedback after 50 trials was intended to incentivise the completion of at least 50 trials per participant. The overall number of trials, and the number of each type of trial seen by each 175 participant was subject to some variation since some participants completed more trials than 176 others and trial type was selected randomly at the beginning of each trial.

177
The lab participants completed 100 trials, and were given the information about their 178 performance relative to others only after they have completed these 100 trials.

179
The task application was coded in HTML and JavaScript with the aid of the CraftyJS

204
The retinal speed of stimuli was subject to variation between participants on the basis of 205 screen size and viewing distance (which were not controlled). This variation is handled 206 statistically as inter-participant variation (Error! Reference source not found.). A more 207 pressing concern is the possibility that the experimental conditions may have been 208 differentially affected by differences in stimulus retinal speed given that one of the conditions 209 (trajectory change) was implemented through a change in the motion of the stimulus. This 210 concern would be apt for manipulations in which the speed of the stimulus was altered (e.g. 211 examining change detection for acceleration), but does not apply when, as here, the speed is 212 kept constant while the direction of motion is altered.

213
The trade-off between these various factors was considered acceptable given the robustness  The excluded trials were examined for differences in manipulation using χ 2 tests with Yates'   .122, all η p 2 < .059).

295
The presence of an interaction between masking and load (Fig 3) indicates that change 296 blindness occurred. The absence of three-way interaction between that interaction and 297 change type is consistent with the suggestion that the change blindness effect is equivalent 298 between change types. These data are consistent with the hypothesis that trajectory changes 299 are detected and missed in a similar manner to orientation changes.