When an edit is an edit can an edit be too much?
The effects of edits on arousal, attention, and memory for television messages.
Shauhua Zhou, Graduate Student
Nancy Schwartz, Graduate Student
Paul D. Bolls, Graduate Student
Robert F. Potter, Doctoral Candidate
Annie Lang, Associate Professor
Gail Trout, Graduate Student
Ruth Funabiki, MA Washington State
Jennifer Borse, Graduate Student
David Dent, Graduate Student
The Institute for Communication Research
Department of Telecommunication
515 N. Park Ave.
Bloomington, IN 47408-3829
Address correspondance to:
Annie Lang ([log in to unmask])
(812) 855-5824, at the above address.
March 31, 1997
Paper submitted to the Theory and Methodology Division of AEJMC.Abstract
This study examines the effect of the number of edits in a television
viewers' arousal and memory for messages. Messages were chosen with 4 levels of
(defined as a cut from one camera to another in the same visual scene): slow
(0-7), medium (8-15 ), fast (16-23), and very fast (more than 24) in 60 seconds.
Frequency and amplitude of
skin conductance responses was used to measure arousal. Percent correct, and
response on a visual recognition task measured memory. Results edits: 1)
arousal, and; 2) increase memory to a point after which it decreases. Results
messages produced with between 8 and 23 edits per minute may maximize memory and
attention and arousal.
When an edit is an edit can an edit be too much?
The effects of edits on arousal, attention, and memory for television messages.
The literature on television effects frequently complains that although
viewers report high levels of exposure to television, memory for what they have
seen is very
low ( Gunter, 1987). On the other hand, both advertisers and educators desire
memorable video products. Advertisers try to produce television ads that are
that lead to purchase of the advertised product. Educators utilize broadcast
television or video
formats to create unambiguous and memorable programs. Recent research
production variables have an impact on how viewers respond to and remember
messages. This study measures the effects of a particular production feature,
edits, on viewers'
processing and memory.
Individuals respond to television messages with reflexive and cognitive
evolved over millennia. Information processing theory suggests that there are
at least three
processes involved in cognitive processing. The first process is encoding
memory. The second process is storage of the information. And the third
process is retrieval.
The limited capacity approach to television viewing (Lang, 1995; Lang,
Kawahara, Potter, & Dent, 1996) suggests that viewers' information processing
limited, and that increasing the cognitive resources allocated to one of these
decreases resources available to other processes. For instance, if more of a
information processing resources are allocated to encoding, fewer resources are
The limited capacity approach to information processing suggests that human
to televised messages are directed both by characteristics of the message and
Production qualities of a television message, such as visual complexity and
some of the responses that limit or enhance processing capacity. Individual
control information processing capacity by increasing or decreasing their
vigilance. The limited capacity approach asserts that once information
processing capacity is
exceeded, some information will not be fully processed. In this case,
individuals may be unable
to recall or even to recognize some message content.
Theorists, such as Craik and Lockhart (1972), Lang (1995), and Zechmeister
Nyberg (1982) suggest that different kinds of retrieval tasks can be used to
completeness of the various sub-processes underlying information processing.
Lang (1995) suggests that free recall tasks can be used to evaluate the effects
of media on the
completeness of information available at the retrieval stage of processing, cued
recall tasks can
be used to evaluate the effects of media on the completeness of the storage
processing, and recognition tasks to evaluate effects of media on the
completeness of the
Research shows that television producers can affect viewers' information
directly by incorporating into their messages production features that affect
the allocation of
processing resources. In particular, many structural features of television
elicit what is called
an orienting response.
The "orienting response" is an involuntary physiological and behavioral
directs our attention toward new or relevant information in the environment.
response is made up of a set of physiological and behavioral responses which
sensory receptors (eyes, ears, nose) toward the stimulus, lowered heart rate,
flow to the muscles, alpha wave suppression in electroencephalogram (EEG),
temperature, increased electrical conductivity of the skin, and increased blood
flow to the brain
Orienting responses to television messages have been measured in various
Several measurement techniques used by psychologists have been adopted by
researchers. Reeves, Thorson, Rothschild, McDonald, Hirsch and Goldstein
among the first to attempt to evaluate attention to television by measuring
decreases in alpha
waves -- called alpha blocking -- in the brain. They found significant alpha
response to movement, camera techniques, and other visual structural features of
(Reeves, Thorson & Schleuder, 1986). Similarly, Lang and her colleagues have
used heart rate
to document orienting responses elicited by: related and unrelated cuts (Lang,
Strickwerda, & Sumner, 1993); movement, cuts, and edits (Lang , 1990);
(Thorson & Lang, 1992); and the onset of negative video (Lang, Newhagen, &
The limited capacity approach to television viewing suggests that when a
message elicits an orienting response, this results in an increase in the
allocation of processing
resources to the message and that this increase in resource alters what viewers
the message. Several studies (Reeves, Thorson, Rothschild, McDonald, Hirsch &
1985; Thorson & Lang, 1989; Lang, 1990; Lang, 1991) demonstrate that formal
features of television (such as cuts, movement, sound changes) which evoke the
response increase attention to television. Both Lang et al. (1993) and Geiger &
(1993) demonstrated that secondary task reaction times (often used as a measure
allocated to processing) increased immediately following related and unrelated
However, despite the fact that orienting responses increase the resources
processing a television message, this does not necessarily result in an increase
in memory for
the message. The effect of the orienting response on memory appears to be
determined by the
cognitive load imposed by the content of the message. For example, Thorson &
demonstrated that memory for videotaped lectures increased following orienting
the content of the lecture was easy for the viewer. However, if the lecture
difficult, memory for information following the formal feature decreased.
Similarly, Lang et al. (1993) looked at memory for information immediately
either related or unrelated cuts. Related cuts were cuts where the information
cut was narratively and semantically related to the information preceding the
cuts occurred when the information following the cut was narratively and
unrelated to the information preceding the cut. Results showed that memory for
following the cut increased for related cuts and decreased for unrelated cuts.
The limited capacity approach suggests that unrelated cuts add more new
to the message than unrelated cuts do. The work of encoding and storing that
information requires more effort or processing resources than encoding and
expected or related information which follows a related cut.
The limited capacity approach predicts this phenomenon because it suggests
elicit orienting which results in more resources being allocated to processing
However, cuts also introduce new information which requires these additional
resources to be
processed. According to this theory, memory for the message will begin to
decline at the point
at which viewers no longer have sufficient processing resources to encode,
process, and store
all the information contained in a message.
A recent study (Lang et al., 1996) extended this work by examining the
effects of the
number of cuts in a message on overall memory for the message. In this study,
defined as a shift from one visual scene to a completely new visual scene.
Thus, a cut always
added new information in the form of a new visual scene. The number of cuts in
messages was varied and memory for the messages was measured. The limited
approach predicts an inverted U-shaped curve since initially cuts will help
viewers to process
the message by increasing the processing resources allocated to the message.
eventually, the viewer will be unable to keep up with the demand for processing
memory will suffer. Results showed that as the number of cuts increased from
slow (0-1 cuts
in 30 seconds) to medium (5-6 cuts in 30 seconds) viewers memory for the
increased. However, as predicted, memory for fast messages (those with 10 or
more cuts in 30
These results suggest, however, that a different story might be told if one
number of related cuts in a message. When examined singly, memory for
information in a
message increased following related cuts (Lang et al., 1993). And because
related cuts add
less "new" information to a message than unrelated cuts, it seems logical to
adding more and more related cuts to a message might not overload a viewers
system as quickly or perhaps at all.
This study was designed to examine that question. For this study cuts are
defined as a
change from one visual scene to another (like the unrelated cuts in previous
work). An edit,
on the other hand, was defined as a change of one camera shot to another within
visual scene. An establishing shot used as a definition of a visual scene and
juxtaposition of shots taken from that same visual scene. In other words, edits
changes within the context of a single location. For example, alternating
during a conversation, would be described as edits. Unlike a cut which takes the
viewer to a
completely new environment, and therefore increases the amount of new
information to be
processed, an edit should elicit orienting but introduce less new information,
require less effort to process. As a result, memory for the message may
increase as a result of
increasing the number of edits in a message, since overload is less likely to
occur. This leads to
Hypothesis 1: As the number of edits in a television message increases,
memory for the
content of the message is likely to increase.
Previous research on the rates of cuts and edits suggests that fast paced
(those with many structural features) elicit, in addition to increased
attention, increased arousal
in viewers. Because arousal is often associated with an increase in memory and
messages, this is an important variable to examine. Increasing the number of
cuts in a message
was clearly shown to increase both viewers self-reports of arousal and their
(measured by skin conductance) in the study reported above (Lang, et al., 1996).
expected that increasing edits will similarly increase viewers' levels of
system activation -- or arousal. Hence:
Hypothesis 2: As the level of edits in a television message increases,
This experiment is a mixed 4 (Order of Presentation) X 4 (Edits) X 5
design. To construct the stimulus tapes, 20 messages, each one-minute long,
from a pool of television programs, advertisements, and feature movies.
chosen to create four levels of edits (slow, medium, fast and very fast). Four
orders were constructed randomly with the aid of a computer software program
Superlab. Order of presentation was the only between subject variable.
Edits was operationalized as a change from one camera shot to another in
scene. A visual scene was defined as the location where related activities took
place. Thus a
football field was a visual scene, the juxtaposition of all shots taken from
which were defined
as edits. But things happening outside the football field and the juxtaposition
of those shots
were considered to be cuts rather than edits.
The four levels of edits were defined according to the following criteria.
within the one-minute message is slow, 8-15 is medium, 16-23 is fast, more than
24 is very fast
paced. All the messages contained fewer than 3 cuts (a level defined as "slow"
Thirty nine communication majors participated in the experiment for class
Visual recognition, heart-rate, and skin conductance were collected for all 39
Memory was measured using a visual recognition task. Snippets of video were
to the subjects who pushed the buttons on a joystick to indicate whether they
snippet or not. Each snippet was about 6 frames long. Subjects viewed 120
separated by 2 seconds of black. Sixty of the snippets came from the messages
seen before while the other half, called foils, came from messages they had not
Both percentage of correct response and the response latency, or the length of
time it took the
subject to respond were measured.
Arousal was operationalized using skin conductance measures. Skin
was collected using two Beckman standard AG-AGCL electrodes place on the palmar
of the subject's non-dominant hand. Data were collected using a Coulbourne skin
coupler connected to a 386 computer outfitted with a labmaster A to D, D to A
conductance was sampled 10 times per second. Score XY a subprogram of VPM was
score spontaneous skin conductance responses (SCR) during each message.
Frequency of skin
conductance responses, and size of the largest skin conductance response were
indicators of arousal (Martin & Venables, 1983; Hopkins & Fletcher, 1994).
An experimenter greeted the subjects upon their arrival. The experimenter
least 15 minutes before the subject in order to set up the lab and run safety
were tested individually. The experimented explained to the subject that his or
her heart rate
and skin conductance data would be recorded using small sensors attached to his
forearms and hands. The subjects signed a consent form before the experiment.
The experimenter stressed that because of the sensitivity of the recording
it was imperative that the subject remain as still as possible during the
experimenter then asked to subject to pay close attention to the stimulus
messages because his
or her memory would be tested later. The experimenter then started the stimulus
tape and the
subject viewed the messages. After the messages, the experimenter instructed the
how to use the joystick. A recognition tape was then played and the subject
they had seen the snippets or not.
This experiment was done in conjunction with two other experiments
elsewhere. As a result subjects performed six tasks during this experiment.
either viewed television messages (this experiment) or read headlines on a
then they listened to a 6 minute radio message, and then they performed the task
done first (i.e. viewed television or read headlines). Following the three
subjects performed recognition tests for all three experiments in the order they
saw the stimuli.
The whole experiment lasted about 1 + hours.
This hypothesis predicted that as the level of edits increased, memory
Significant main effects were found for both latency (F(3, 93)=28.90, p<.0001
epsilon-squared=.3544), shown in Figure 1, and percent correct responses (F(3,
epsilon-squared=.2736), shown in Figure 2. Interestingly, both the latency and
correct data fail to support the hypothesis. Rather, they show, as was the case
that increasing the level of edits improves memory up to a point (somewhere
and fast) after which recognition memory (indicative of thoroughness of
suggesting viewers processing systems are unable to keep up with the message.
correct was highest for medium level of edits (91.9 percent) followed by fast
and very fast (84.4 percent), slow paced messages yielded the lowest percentage
answers (78.4 percent).
On the other hand, latency measures of memory showed the fastest response
(indicative of the best memory) for medium (M=866.69), slow (M=909.53), very
(M=948.23), and fast (M=1203.95).
This hypothesis predicts that viewers' arousal increases as the level of
The main effect for both number of skin conductance responses (F(3, 75)=13.80,
epsilon-squared=.3299) and the size of the largest response (F(3, 75)=6.98
p<.0001, epsilon-square= .1869) were significant. As predicted, the frequency
of responses (shown in Figure 3)
was much higher in the fast paced messages (M=3.12 for fast messages; M=3.01 for
messages) than the slow (M=2.02) and medium paced messages (M=2.08). The size of
response data showed a similar pattern (shown in Figure 4). Fast paced messages
larger responses (M=1.16 for fast messages; M=.97 for very fast messages) than
and medium paced messages (M=.78).
Results of this experimental study suggests 3 things. First, as predicted,
number of edits in a television messages increases viewers' level of autonomic
Second, increasing the number of edits in a message increases viewers'
recognition memory for
the message to a point, after which it decreases -- the same pattern that was
shown in previous
research for cuts. Third, by increasing the number of edits in a message,
increase viewers' attention and arousal levels. However, to maximize memory for
message, producers' must choose an appropriate level of number of edits
somewhere between 8 and 15 edits per minute).
These results may challenge the notion that edits might not introduce
information to ever overload viewer's information processing capacity (Bolls,
Hibbs & Lang,
1995). Rather, these results suggest that edits are as likely to overload
processing as cuts.
Further, they appear to overload processing at approximately the same rate that
cuts do. In
previous research, evidence of decreasing memory in response to increasing cuts
was found at
about 7-10 cuts per 30 seconds. This study showed decreasing memory at between
16 and 23
edits per minute, which would correspond to 8-12 edits per 30 seconds. Thus,
differences in amount of information being introduced by the two structural
features there is
little evidence that this affects what is processed and remembered from the
It is also interesting to note that a medium level of edits is much better,
than a slow level of edits. It would be interesting for future research to
precisely where the threshold lies as this may have important practical
implications for the
industry interested in getting viewers' attention and maximizing viewers'
These results also show that, like cuts, edits can elicit autonomic arousal
Since arousal affects most aspects of information processing as well as liking
for messages, it is
good to know that simply using a two camera rather than a one camera production
increase arousal. It is interesting to note that the pattern of these results
suggests that the
relationship between arousal and edits may not be precisely linear. Slow and
medium levels of
edits appear to result in a low level of arousal while fast and very fast result
in a higher level of
arousal. Suggesting more of a step function than a linear relationship. This
is different from
previous research using cuts which showed increasing arousal levels for each
level of cuts.
However, this previous research, unlike this study, controlled for the
arousingness of the
content of the messages. Future research to further investigate the shape of
between edits and arousal should control for arousingness of message content.
Finally the response latency results present an interesting question. Why
is the response
latency for very fast paced messages faster than the response latency for fast
Theoretically, messages with the most edits should have elicited the longest
One possibility is that for very fast messages less of the information is from
the message may
be encoded, but (because the messages are highly arousing and receive high
attention) what is
encoded is stored extremely well. Thus, even though only a tiny part of the
processed and encoded, as Cameron and Fresco (1994) would argue, response
this situation is short because of a strong, but limited, set of memory traces
which are rapidly
This explanation might suggest that arousal (rather than level of edits)
may be a better
indicator of latency of response. Certainly visual inspection of these data
suggest that level of
autonomic arousal appears to be a good predictor of the latency of response
given by the
In sum, these results continue to support the usefulness of the limited
approach to television viewing as a framework for investigating the impact of
variables on the information processing television messages. Future research
to further explicate the effects of edits and other structural features of
television on viewers'
processing of messages.
Figure 1: Latency of Response by Edits
Figure 2: Percent correct by Edits
Figure 3: Frequency of SCR by Edits
Figure 4: Size of largest SCR by Edits References
Bolls, P., Hibbs, H., & Lang, A. (1995). The medium is the memory: Using
features to predict memory for random television messages . Presented to the
Systems Division of the International Communication Association annual
Albequerque, NM. May.
Cameron, G. F. (1994). Response Latency. In A. Lang (Ed.), Measuring
Psychological Responses to Media : Lawerence Earlbaum Associates.
Geiger, S., & Newhagen, J. (1993). Revealing the black box:
processing and media effects. Journal of Communication, 43(4), 43-50.
Geiger, S., & Reeves, B. (1993). The effects of scene changes and semantic
on attention to television. Communication Research, 20, 155-175.
Gunter, B. (1987). Poor Reception: Misunderstanding and forgetting
Hillsdale, NJ: Lawrence Erlbaum Associates.
Hopkins, F. (1994). Skin Conductance. In A. Lang (Ed.), Measuring
Responses to Media : Lawerence Earlbaum Associates.
Kawahara, K., Bolls, P., Hansell, R., Potter, R., Dent, D., & Lang, A.
effects of production pacing and content arousal on viewers allocation of
capacity to encoding
and storage of television messages . Paper presented to the International
Association, Information Systems Division, Chicago, IL: May.
Lang, A. (1991). Emotion, formal features, and memory for televised
advertisements. In F. Biocca (Ed.), Television and political advertising (Vol.
Vol. I, pp. 221-244). Hillsdale, NJ: Lawrence Erlbaum, Associates.
Lang, A. (Ed.). (1994). measuring psychological responses to media:
Lang, A., Bolls, P.D., Kawahara, K., Potter, R. & Dent, D. (1997). The
production pace and arousing content on the information processing of television
Lang, A., Dhillon, K., & Dong, Q. (1995). The effects of emotional arousal
valence on television viewers' cognitive capacity and memory. Journal of
Electronic Media, 39(3), 313-327.
Lang, A., & Friestad, M. (1993). Emotion, hemispheric specialization, and
redundancy. Communication Research, 20(5), 647-670.
Lang, A., Geiger, S., Strickwerda, M., & Sumner, J. (1993). The effects of
unrelated cuts on viewers' memory for television: A limited capacity theory of
viewing. Communication Research, 20, 4-29.
Lang, P. J., Greenwald, M., Bradley, M. M., & Hamm, A. O. (1993). Looking
pictures: Evaluative, facial, visceral, and behavioral responses.
Psychophysiology, 30, 261-273.
Lynn, R. (1966). Attention, arousal, and the orientation reaction. Oxford:
Martin, Veenables, (1983). Handbook of Psychophysiological Recording.
Reeves, B., Thorson, E., & Schleuder, J. (1986). Attention to television:
theories and chronometric measures. In J. Bryant & D. Zillmann (Eds.),
Perspectives on Media
Effects . Hillsdale, NJ: Lawrence Erlbaum, Associates.
Reeves, B., Thorson, E., Rothschild, M., McDonald, D., Hirsch, J., &
(1985). Attention to television: Intrastimulus effects of movement and scene
changes on alpha
variation over time. International Journal of Neuroscience, 25, 241-255.
Thorson, E., & Friestad, M. (1989). The effects of emotion on episodic
television commercials. In P. Cafferata & A. M. Tybout (Eds.), Cognitive and
Responses to Advertising (pp. 305-326). Lexington, MA: Lexington Books.
Thorson, E., & Lang, A. (1992). The effects of television videographics and
familiarity on adult cardiac orienting responses and memory. Communication
Zechmeister, E. B., & Nyberg, S. E. (1982). Human memory, An introduction
research and theory. Monterey, CA: Brooks/Cole Publishing Company.