Chapter 3: Hiding a Cut
3.3 Directing attention internally
3.3.3 Positioning a cut relative to an event
The evidence from event segmentation and disruption blindness previously cited leads to several conclusions:
1. Breakpoints are important for the successful perception, recall, and recognition of events (Newtson & Enqguist, 1976; Schwan & Garsoffky, 2004; Zacks & Tversky, 2001);
75 2. Attention is occupied with the encoding and storage of an event leading up to its
expected end (Baird & Baldwin, 2001; Carroll & Bever, 1976; Schwan et al., 2000; Smith et al., in press; Whitwell, 2005);
3. For a 400ms period after a breakpoint attention is susceptible to capture by visual disruptions (Baird & Baldwin, 2001; Newtson & Enqguist, 1976; Saylor & Baldwin, 2005; Smith et al., in press; Whitwell, 2005).
Is it possible, based on these conclusions to suggest where, relative to a breakpoint, the best place to position a cut would be so that the viewer is not aware of the cut? There are four possible locations of a cut: during an event (During), leading up to the end of an event (End), at the breakpoint between events (Breakpoint), or at the beginning of a new event (Beginning). The suitability of all of these will be addressed in turn.
Cutting during an event means identifying time points not adjacent to breakpoints. As breakpoints have been seen to correlate with periods of increased motion (Newtson et al., 1977; Zacks, 2004) time points in the middle of events are highly likely to involve an absence of visual motion i.e. stasis. These periods are also characterised by their insignificance to the successful perception, recognition, and recall of the overall event (Newtson & Enqguist, 1976; Schwan & Garsoffky, 2004; Zacks & Tversky, 2001). This means that there is probably no reason for attention to be allocated to encoding and conscious perception during these periods (Levin & Varakin, 2004). When attention is unfocussed it is vulnerable to capture by visual disruptions such as the whole-field disruptions caused by a cut (Simons, 2000). This is exactly as has been observed in disruption blindness experiments: visual disruptions ranging from 200-600ms motion-fields or blanks, down to single sepia or blank frames inserted during periods of stasis are perceived more often than when presented during motion (Levin & Varakin, 2004; Saylor & Baldwin, 2005). Without attention being internally occupied the sudden changes to the visual scene caused by
76 the cut, such as apparent motion of objects, is highly likely to capture attention (Folk et al., 1994).
There are also perceptual consequences of a cut during a period of visual stasis. Lang, Geiger, and colleagues have shown that all cuts have the potential to trigger an orienting response (Geiger & Reeves, 1993; Lang et al., 1993). This increases cognitive resources which, if the viewer is occupied with a cognitive task, will improve performance on the task (Lang, 2000). If the viewer is not occupied, such as when watching an insignificant part of an event, these extra cognitive resources will not result in the improved encoding of the visual event (Schwan et al., 2000). This indicates that a cut during an insignificant part of an event is not perceived as indicating a breakpoint in the event (Schwan et al., 2000). Instead the extra cognitive resources lead to increased awareness of the cut (d'Ydewalle & Vanderbeeken, 1990; Schröder, 1990).
In conclusion, all available evidence seems to suggest that cutting during an event when the visual scene is static will not hide a cut.
Placing a cut at the end of an event is better than during the event as the occurrence of visual motion is much higher (Newtson et al., 1977; Zacks, 2004). Cutting during motion has its advantages as the primary motion will attract attention and lessen the chance that viewers will become aware of a secondary visual disruptions such as a cut (Levin & Varakin, 2004; Saylor & Baldwin, 2005). However, if this secondary disruption interrupts the primary motion by making it jump ahead in time or space or pause momentarily, viewers may become aware of it (Baldwin et al., 2001; Newtson & Enqguist, 1976). To protect against this the moving object would have to be collocated across the cut and follow a continuous path (i.e. be graphically matched across the cut).
77 Deciding exactly when to cut prior to the breakpoint is very important due to the period of cognitive overload associated with the orienting response (OR). After every OR there is at least a 150ms period during which visual information cannot be processed (Geiger & Reeves, 1993). This increases to 300ms when the visual disruption is large (Geiger & Reeves, 1993). After this period of cognitive overload, extra cognitive resources allow for the increased encoding of new visual information but any information presented during the period of overload is lost (Lang, 2000). If the cut occurred less than 300ms (~ 8 frames) before a breakpoint, the visual information associated with the breakpoint would not be perceived. Given the importance of this information for the successful perception, recognition, and recall of the overall event (Newtson & Enqguist, 1976; Schwan & Garsoffky, 2004; Zacks & Tversky, 2001) a cut occurring in this position could lead to an inability to accurately comprehend the event.
In conclusion, cutting immediately prior to a breakpoint may stop the viewer becoming aware of the cut (if object motion and location is preserved across the cut) but there is the risk that it will have negative effects on comprehension of the overall event.
A breakpoint is a transition from one event to another. Whilst the time around a breakpoint may be related to a high degree of visual motion (Newtson et al., 1977; Zacks, 2004) the breakpoint itself is probably a momentary point of stasis as one motion ends and another begins. The absence of any movement within the event to attract the viewer’s attention could explain why the visual transients associated with a cut are ore likely to capture attention during a breakpoint compared to periods either side (Baird & Baldwin, 2001; Saylor & Baldwin, 2005). Cognitive processing of the previous event should also be coming to an end, as indicated by pupil contraction (Smith et al., in press; Whitwell, 2005), meaning that any extra cognitive resources made available by the OR may not benefit encoding of the previous event. However, there is evidence that when cuts coincide with breakpoints viewers’ recall of breakpoints improves significantly compared with presentation of the breakpoints
78 without a cut (Schwan et al., 2000). This indicates that the increase in cognitive resources34 is used to encode the breakpoint. However, this does not rule out the possibility that the attention capture caused by the cut does not also lead to awareness of the cut.
There is also another potential drawback associated with cutting at breakpoints. If an OR is triggered during the moment of stasis between events, by the time perceptual sensitivity has returned 150-300ms later the key change from stasis to motion signifying the beginning of the new event might have been missed. This might create an intact perceptual representation of the event prior to the cut but an incomplete representation of the new event. Critically, the transition connecting the two events might not be perceived. The importance of this transition for the perception of spatiotemporal continuity will be discussed in chapter 5.
In conclusion, the momentary stasis associated with a breakpoint increases the likelihood that a cut captures attention. However, the extra cognitive resources created by the OR improve recall of the breakpoint. What is not known is if the viewer is also aware of the cut and if the immediate cognitive overload associated with the OR hinders their perception of the connection between the old and new events.
Cutting at the beginning of a new event benefits from the cut being hidden in visual motion (see 184.108.40.206), encoding of the previous breakpoint already being completed (Smith et al., in press; Whitwell, 2005), and the transition between the old and the new event already perceived. The OR triggered by a cut in this location might not lead to improved processing of the previous breakpoint but this is already recalled accurately without the co-occurrence of a cut (Newtson & Enqguist, 1976; Schwan et al., 2000). If a cut occurs at the beginning of a new event the viewer will be occupied
79 with processing the transition of between the old and new events. This will be what benefits from the increased resources triggered by the cut. Also, the period of visual motion not perceived due to cognitive overload could be “filled-in” based on the start of the motion perceived before the cut and the motion perceived after the overload35. Therefore, whilst cutting at a breakpoint improves recall of the breakpoint, cutting at the beginning of a new event ensures that the new and old events are fully processed whilst also hiding the visual disruption of the cut in motion.