Exit

In document An attentional theory of continuity editing (Page 168-171)

Chapter 4: Cuing a Cut

4.4 Experiment 1: Discussion

4.4.1 Exit

The results of this study indicate that subjects expect the principal object to fully leave the screen before a cut (100% Exit). Both “occlusion” hypotheses are supported by these results. 100% Exit is expected even if the cut is unpredictable

82 An attentional cue that shifts a person’s attention away from the cue itself. The other main category of cues. pull cues attract attention to themselves.

168 (occlusion expectation hypothesis) and subjects adapt most successfully to 100% Exit (occlusion advantage hypothesis).

The occlusion expectation hypothesis predicted that when the cut was unpredictable subjects would default to expecting a cut to occur once the principal object was fully occluded. This was supported by the results. The pattern of correct response rates prior to the cut indicated that subjects were employing one viewing pattern across all three Exit conditions. Subjects waited until the object was half occluded then began preparing the saccade. This was evident in the significant decrease in correct response rates at cue position -3 in the 100% Exit condition. The total duration of the period of attention withdrawal was 208ms. When the cut occurred before the focal- object was fully occluded the saccade was performed in response to the cut. This resulted in periods of attention withdrawal 208-250ms in length when the cut occurred at 0% Exit and 125ms at 50% Exit. 125ms is insufficient time to prepare and carry out a voluntary saccade (Rose et al., 2002) indicating that the saccades performed in response to a cut at 50% Exit may have been reflexive, i.e. not due to deliberate control by the subject.

The preference for 100% Exit was also observed during analysis of the occlusion advantage hypothesis. As predicted, even when 0% and 50% Exit were consistent across all cuts (within Exit subject groups) signs of attention withdrawal prior to the cut were only seen in the 100% Exit group. The 100% Exit group recovered full attention after only 82-125ms. 50% Exit also showed a very rapid recovery back to pre-cut accuracy (125ms after the cut) although this was in response to the cut. This evidence of short duration saccades further suggests that under some conditions saccades were reflexive.

When the effect of learning was examined all Exit groups showed some improvement in correct response rates over the course of the experiment but the differences were only significant for 100% Exit (cue position 1) and 50% (cue position 3). Reaction times were also found to be quicker for 100% and 50% Exit at the end of the experiment although the difference was only significant for 100% Exit.

169 Both the occlusion expectation and advantage hypothesis are supported by the results of this study but there is not the clear advantage for 100% Exit that was predicted. Subjects only appear to be able to anticipate a cut when it occurs after the focal- object is fully occluded. This anticipation is seen in the withdrawal of attention associated with a saccade to the opposite screen edge. As the period of attention withdrawal starts before the cut it suggest that subjects are “blind” to a 100% Exit cut. This supports the matched-exit/entrance convention of using occlusion of the focal-object by the screen edge to hide a cut. However, the precise convention used in matched-exit/entrance editing is for the focal-object to be 50% occluded before the cut (4.1.2.2). The results of this study suggest that such a cut would not be anticipated by the viewers. As 50% Exit shows no sign of anticipatory attention withdrawal the viewers should remain aware of the cut. However, the rapid recovery of attention after the cut observed for 50% Exit suggests that the saccadic eye movement performed in response to the cut may be reflexive.

Reflexive saccades do not require preparation as they are usually the result of attentional capture. This leads to saccades with significantly shorter durations than if the saccade was performed voluntarily (Sheliga, Riggio, & Rizzolatti, 1995; Walker, Walker, Husain, & Kennard, 2000). Reflexive saccades under the 50% and 100% Exit conditions could either be a result of the edit screen edge pushing attention across the screen in the same way an observed gaze-shift does, or being pulled by the sudden appearance of the focal-object on the other side of the screen. Attention could only be pulled across the screen if the focal-object was visible immediately after the cut. Therefore, if attention is being pulled there should be a clear difference in the duration of attention withdrawal across Entry conditions. This possibility will be discussed in reference to the pursuit initiation hypothesis.

Even though 50% Exit shows rapid recovery of attention after the cut there could still be an advantage for 100% Exit. The occlusion hypotheses were based on the assumption that full occlusion before disappearance would lead to continued perception of the focal-object across the cut (Gibson et al., 1969). Such continuity of

170 perception should not occur when attention is captured by the object’s relocation (e.g. under 50% Exit) as such sudden object relocation will violate spatiotemporal continuity. Therefore, 100% and 50% Exit may provide two useful but perceptually different types of cut. If an editor’s intention is to create a cut with perceived continuity of action they may be able to create this effect with 100% Exit. However, if they just want to ensure that their viewer’s attention immediately shifts to the focal-object of the new shot they can use 50% Exit to capture attention. When presenting editing conventions such as those associated with matched-exit/entrance cuts editors always attach a disclaimer stating that the conventions will vary depending on the action depicted and the editor’s intention. To be able to identify exactly how editor’s use occlusion by the screen edge to cue cuts a survey of matched-exit/entrance cuts would need to be performed and their perceptual consequences judged. Such a survey is beyond the scope of this thesis. Instead the perceptual consequences of such cuts will be discussed in chapter 5.

In terms of the main question of this chapter: Does a viewer expect a cut to occur as the focal-object is fully occluded by the screen edge?, the results related to Exit indicate that viewers do expect a cut only after full occlusion. There also seems to be some advantage of 50% Exit but as this appears to be due to involuntary attentional orienting it can not be due to viewers expecting a cut at 50% Exit. The clearest result indicated by analysis of Exit conditions is that 0% Exit is by far the worst place to cut. A cut when the focal-object is still on the screen, irrespective of whether the cut is expected or not, results in a extended period of decreased performance after the cut, less improvement over repeated presentations, and slower reaction times. Whatever the degree, occlusion by the screen edge creates a less disruptive cut.

In document An attentional theory of continuity editing (Page 168-171)