Television, Arousal, and Memory

The Effects of Production Pacing and Arousing Content on Encoding, Storage, and Retrieval of Television Messages

Paul Bolls

Graduate Student
School of Communication
Washington State University
Pullman, WA 99164-2520

09952477@WSUVM1.CSC.WSU.EDU

Robert F. Potter

Graduate Student
Department of Telecommunications
Indiana University
Bloomington, IN 47405

Annie Lang

Associate Professor
Department of Telecommunications
Indiana University
515 N. Park Ave.
Bloomington, IN 47405
(812) 855-5824

ANLANG@INDIANA.EDU

Paper presented to the Seventh Midwest Artificial Intelligence and
Cognitive Science Conference

Introduction

This paper explores how one element of a television message's production (the number of cuts - defined as a sudden change from one visual scene to another in a television message) interacts with the arousingness of the message's content to alter how viewers encode, store, and retrieve the information contained in the message. This paper uses a limited capacity information processing model of television viewing to investigate these effects.

This model defines the television stimulus as two continuous streams of information (one audio and one visual) which are variably redundant. Further, the television message is defined as having structural elements (e.g. light, sound, camera techniques, video graphics, slow-motion, editing techniques, movement, etc) and content elements (e.g. genre, humor, emotion, narrative structure, etc.).

The viewer is conceived of as a limited capacity information processor. Attention is defined as the allocation of limited processing resources to various aspects of a task. Processing resource allocation is defined as having two dimensions. The first is an automatic phasic process related to selection of stimuli for encoding. The second is a controlled tonic process related to mental effort. The first dimension is primarily controlled by elements of the environment or the stimulus (in this case the structure of the television message). The second dimension is primarily controlled by elements of the viewer (in this case, interest, knowledge, effort, etc.)

Viewing television (the task) is defined as the interaction between a message's content and structure and the viewer's information processing system. Viewing television is the task of encoding, processing, and storing the contents of the message. How thoroughly that task will be performed is determined both by the amount of capacity a message requires to be fully encoded, processed, and stored (which is determined both by its structure and its content), and the amount of capacity a viewer actually allocates to the task.

Research using this model has demonstrated that many aspects of television production (sounds, silence, cuts, edits, sudden movement, video graphics, etc.) result in an increase in resources allocated to television viewing through the elicitation of an orienting response (as indexed by cardiac deceleration, increases in skin conductance, and increases in secondary task reaction times) (Lang et al., 1993). Research suggests that the more structural features included in a message the greater the amount of resources allocated to processing the message (Kawahara et. al, 1995). Basil & Lang (1996) suggest a model which predicts that this additional capacity is allocated primarily to the task of encoding the message, recent data (Kawahara et al., 1996) supports this contention. As the number of structural features in a message increases (that is, the pacing of the production is increased), the capacity allocated to encoding increases. This improves memory (in particular recognition) for the contents of the message, up to a point. However, research suggests that there is a point at which the pacing is too fast and there are insufficient resources available to answer the demand for resources, the viewer becomes overloaded and as a result the task (indexed by recognition for the message) is performed less well (Lang, 1995).

This study will manipulate production pacing by controlling the number of cuts (that is changes from one visual scene to another) in a 30 second message. Previous research has demonstrated orienting responses to cuts and increases in secondary task reaction times during cuts (Lang et al, 1993). It is predicted that increasing the pacing of the television message will increase the capacity allocated to encoding the incoming message.

Another result, however, of increasing pacing is to increase viewers' arousal levels (as measured by self-report and skin conductance (Kawahara, et al., 1996)). This model also suggests that arousal plays an important role in resource allocation. In particular, this model hypothesizes that arousing content causes processing resources to be allocated to the process of storing information which has already been encoded from the message. This hypothesis is based on research which shows that the arousingness of a message's content (where the messages might be still slides (Bradley et al., 1992) or television messages (Lang, Dhillon, & Dong, 1995)) improves viewers' free and cued recall for messages but not necessarily their recognition of the messages.

If this is the case, then increases in the production pacing of a message should increase the processing resources allocated to storage (because it elicits arousal in viewers) as well as the processing resources allocated to encoding. In addition to experiencing arousal caused by the increase in production pacing, viewers may also experience increased arousal as a result of a message's content. Thus, this model also predicts that when messages are about arousing topics more processing resources will be allocated to storage.

In the case where messages are both arousing in content and fast paced in production there will be increased calls for capacity to both encoding (as a result of pacing) and to storage (as a result of pacing and arousing content). This means that viewers are likely to become overloaded sooner when viewing arousing messages than when viewing calm messages.

The following study was designed to test this model. Production pacing was manipulated through controlling the number of cuts in a messages. Arousingness of content was controlled by selecting messages which groups of coders rated as being either very calm or very arousing. Encoding was indexed through forced choice recognition measures. Storage was indexed using cued recall measures. Retrieval was indexed using free recall measures.

Methodology

Design and Independent Variables

The experiment is a mixed 3 (Order of Presentation) X 3 (Pacing) X 2 (Arousal) X 5 (Message) design. To construct the stimulus tapes; 30 messages were chosen from a pool of 312 coherent 30 second messages which had been taped off the local cable system (not including premium channels). Two levels of arousal (calm and arousing) and three levels of Pacing (slow, medium, and fast) were completely crossed. Five messages were chosen in each arousing/pacing category, resulting in a total of thirty messages. Three semi-random presentation orders were constructed and order of presentation was the only between subjects variable. Orders were constructed in blocks of six messages. The six messages in each block contained one message from each Arousing/Pacing category. The messages making up each block were randomly chosen for each order with the constraint that, across the three orders, each individual message had to appear in the first or last block of six one time.

Pacing was operationalized as the number of cuts in a 30 second television message. Slow paced messages had 0 or 1 cut, medium paced messages had 4-6 cuts, and fast paced messages had 11 or more cuts in thirty seconds.

Arousingness of content was operationalized in two ways. First, the messages were originally chosen to be high or low in content arousingness. This was done by having at least three undergraduate coders rate the pool of 312 messages using SAM (the Self-Assessment Mannequin) developed by P.J. Lang (Bradley et al., 1992). SAM is a pictorial arousal scale which translates into a 9 point scale ranging from 1=Very aroused or excited to 9=calm, sleepy, not aroused. Arousal was also measured (during the experiment) by having viewers rate their own arousal levels using the same measure to ensure that both the pacing manipulation and the content arousingness manipulation increased viewers' self-reported levels of arousal. Results showed that subjects rated the arousing messages (M=4.97) as significantly more arousing than the calm messages (M=6.62) (F(1,46)=150.00, p<.0001). Viewers also rated fast paced messages (M=5.61) more arousing than medium paced messages (M=6.03) which were more arousing than slow paced messages (6.79) (F(2,92)=47.00, p<.000).

Dependent Variables

The dependent variables were recognition, cued recall, and free recall. Recognition was used to assess encoding and was measured using forced choice, four alternative, multiple choice tests with four questions per message. To control for fatigue effects, the multiple choice tests were arranged in four systematically different orders that were not related to the orders of the stimulus tape. Results are reported as percent correct (25% being the level achievable by chance).

Cued recall was measured by having subjects fill out a 30 page booklet. Each page had the sentence "Write down everything you can remember about" followed by a unique cue for each message. Cues were constructed by naming the major subject of each message. Occasionally a cue contained a second descriptor to clearly differentiate it from a similar message. At this point, these data have only been coded for the number of words written per message (a fairly gross measure of cued recall).

Free recall for the entire message was used to assess retrieval. To measure free recall participants were given a single sheet of paper with 30 numbered blank lines on it. They were instructed to write a brief sentence about each message they could remember viewing.

Participants

Ninety-six undergraduates at a large Western University participated in this experiment for extra credit in a Communications course. Participants were assigned to one of two memory conditions. There were 47 subjects in the recognition/free recall condition and 49 subjects in the cued recall condition.

Procedure

Participants viewed the stimulus tape in groups of 2-6 on a 19 inch color television. Data for this experiment was collected in two sessions. Participants in the first session were given the free recall task followed by the recognition task. Participants in the second session were given only the cued recall task. After viewing the stimulus tape and completing the appropriate memory tasks, participants were thanked and dismissed.

Hypotheses and Results

The model used here predicts different amounts of processing resources will be allocated to the sub-processes of encoding and storage as a result of varying the levels of pacing and arousal in television messages. These predictions and the results will be presented first, for the recognition data (or the encoding process), second, for the cued recall data (the storage process), and finally for the free recall data (which represents retrieval and is thought to be primarily dependent on storage).

Pacing, Arousal, and Encoding

The model predicts that as the production pacing of a message increases the allocation of resources to encoding will increase and that eventually there will be insufficient resources to respond to additional calls for resources at encoding and performance on the encoding task will fail. Therefore, as Pacing increases, viewers should encode more and therefore recognize more of the specific content of a message up to the point where capacity is overloaded, at which point they should encode less (or recognize less) of the specific content of a message. Thus:

H1: There will be a main effect for Pacing on the recognition data such that as pacing increases recognition for the messages content will increase up to a point, after which it will level off or actually decrease.

This hypothesis was supported. The main effect for Pacing on the recognition data was (F(2,92)= 5.54, p<.0053). The means are shown in Table 1. As predicted medium messages are remembered better than slow messages and fast message are remembered slightly less well than medium messages.

Table 1: Mean Recognition Scores by Pacing and Arousal

Pacing      Calm Messages  Arousing Messages   All Messages

Slow Messages     45.95a         57.49b         51.72

Medium Messages   59.14b         54.32c         56.73

Fast Messages     62.55b         46.37          54.46

All Messages      55.88          52.73

The model goes on to suggest that arousingness of content should increase capacity allocated to storage, it does not directly suggest that the capacity allocated to encoding will increase. However, some previous research suggests that recognition for arousing television messages will be higher than recognition for calm messages because arousal is also associated with greater liking for the messages (Lang et al., 1995; Kawahara et al., 1995). Therefore:

H2: It is predicted that there will be a main effect for Arousingness of Content on the recognition data such that Arousing messages will be recognized better than calm messages.

This hypothesis was not supported. While the main effect was significant (F(1,46)=6.21, p<.0164), the means (shown in Table 1) are in the opposite direction with calm messages being recognized better than arousing messages.

Finally, the model suggests that the point at which there are insufficient resources to answer the demands for encoding made by increased pacing will occur sooner for arousing messages than for calm messages because of the additional need for capacity at storage caused by the message's arousing content.

H3: There will be a Pacing by Arousingness of Content interaction such that, for calm messages increased pacing will result in steadily increasing recognition with a possible leveling off of the increase at high levels of pacing. For arousing messages, however, recognition should increase from slow to medium paced, but is likely to actually decrease significantly for fast paced messages.

This hypothesis was also supported (F(2,92)=56.21, p<.0001). The means are shown in Table 1. As predicted calm messages show steady increases in recognition as pacing increases, but arousing messages show a significant drop off in recognition at high levels of pacing.

Pacing, Arousal, and Storage

The major prediction of the model related to the storage process, is that arousal experienced by the viewer, caused either by the content or the pacing of the message, will result in additional capacity being allocated to storing the message. As a result, both increases in pacing and the arousingness of message content should increase the capacity allocated to storage and therefore should increase viewers' cued recall for the messages. When messages are both fast paced and arousing there will be the greatest call for processing resources at both the encoding and the storage stages, which should result in the greatest overload of a viewers' processing system. Thus:

H4: There will be a main effect for Pacing on the cued recall data such that as pacing increases cued recall for the message will increase.

The main effect for Pacing was significant (F(2,96)=14.01, p<.000). The means (shown in Table 2) demonstrate that cued recall increased to a point and then levels off.

Table 2: Cued Recall scores by Pacing and Arousingness of Content

Pacing      Calm Messages  Arousing Messages   All Messages

Slow Messages     9.25c          10.96a         10.11a

Medium Messages   10.38d         15.08b         12.73b

Fast Messages     13.21a         11.81a         12.51b

All Messages      10.95          12.62

H5. There will be a main effect for Arousingness of Content on the cued recall data such that arousing messages will show higher cued recall scores than calm messages.

This hypothesis was supported (F(1,48)=15.14, p<.000), the means are shown in Table 2. As predicted arousing messages have higher cued recall scores than calm messages.

H6: There will be an interaction between Pacing and Arousingness of Content such that, for calm messages, as pacing increases cued recall will increase. On the other hand, for arousing messages, as pacing increases cued recall will increase to a point, and then decline.

This hypothesis was also supported (F(2,96)=14.01, p<.000). The means (shown in Table 2) demonstrate steadily increasing cued recall for Calm messages but an inverted-U for arousing messages, with cued recall falling off sharply for fast paced messages.

Pacing, Arousal, and Free Recall

Finally, this experiment also assessed free recall. While no specific relationships between capacity allocation during viewing and retrieval of the information post-viewing were made, it is logical to suppose that messages which are encoded and stored better will be retrieved better. On the other hand, messages which overload the viewers' processing system may also be retrieved quite well. After all, viewers are working as hard as they can to process these messages. Thus, while the amount of specific information which viewers are able to encode and store may decrease -- it is logical to predict that what they do manage to encode and store may be quite retrievable. Since the free recall measure used here does not measure amount of memory (that is how much of the information contained in the message was remembered) but only if the message was remembered, it is likely that retrieval will be more related to how well a message was stored (and therefore to arousal levels) than to encoding. Therefore:

H7: There will be a main effect for Arousingness of Content on the free recall data such that arousing messages are remembered better than calm messages.

This hypothesis was not supported (F(1,46)=.15, p<.70). The means are shown in Table 3.

Table 3: Free recall scores by Pacing and Arousingness of Content

Pacing      Calm Messages  Arousing Messages   All Messages

Slow Messages     36.33          43.67          40.00

Medium Messages   51.05          37.55          44.29

Fast Messages     43.27          51.84          47.55

All Messages      43.54          44.35

H8: There will be main effect for Pacing on the free recall data such that as pacing increases, free recall for the message will increase.

This hypothesis was supported (F(2,92)=4.51, p<.0135). The means (shown in Table 3) demonstrate that free recall for a message increases steadily as pacing increases.

H9: There will be an interaction between Pacing and Arousal such that, as pacing increases free recall will increase and the effect will be greater for arousing messages.

This interaction was significant (F(2,92)=6.35, p<.0026). The means (shown in Table 3) generally support the hypothesis. Fast messages are free recalled better than slow messages for both calm and arousing messages. But, for medium paced messages, calm medium are the best recalled of the calm messages, while fast medium are the worst recalled of the arousing messages.

Discussion

This study tested predictions made about how viewers encode, store, and retrieve television messages based on a limited capacity information processing model of television viewing. By and large, the hypotheses supported the model. In particular, hypotheses derived from the premise that capacity is allocated not globally to the task of television viewing but rather separately to the sub-processes of encoding and storage were upheld. Further support was also provided for the separation of structure and content variables both in terms of memory and in terms of how they might affect processing resource allocation.

References

Bradley, M., Greenwald, M., Petry, M. & Lang, P. (1992). Remembering Pictures: Pleasure and arousal in memory. Journal of Experimental Psychology, 18(2), 379-390.

Kawahara, K., Bolls, P., Hansell, R., & Lang, A., Potter, R. F. & Dent, D. R. (1996). The effects of pacing and Arousingness on Processing Resource Allocation. Paper to be presented to the InternationalCommunication Association, Chicago, IL.

Kawahara, K., Wadleigh, P.M., Hansell, R., Hazel, M.T., Nagami, K., and Lang, A. (1995). Everybody loves a fast message! Pacing, arousal and memory for television messages. Paper presented to the Association for Education in Journalism and Mass Communication, Wash. DC.

Lang, A. & Basil, M. D. (1996). What do secondary task reaction times measure anyway? Paper to be presented to the International Communication Association, Chicago, IL. May.

Lang, A. Dhillon, P., and Dong, Q. (1995). Arousal, Emotion, and Memory for television messages. Journal of Broadcasting and Electronic Media, 38,1-15.

Lang, A. (1995). Defining audio/video redundancy from a limited capacity information processing perspective. Communication Research, 22, 86-115.

Lang, A., Geiger, S., Strickwerda, M., Sumner, J., (1993) The effects of related and unrelated cuts on viewers memory for television: A limited capacity theory of television viewing. Communication Research, Vol. 20; 1. pp. 4-29.