The Emotional Significance of Color in Television Presentations

 

Benjamin H. Detenber

Nanyang Technological University

Robert F. Simons and Jason E. Reiss

University of Delaware

 

Contents

Abstract

A within-subjects experiment was conducted to investigate the emotional effects of color in film and television clips.  The study involved obtaining physiological measures (skin conductance, heart rate, and facial muscle movement) during the presentation of 54 short (6 s) film clips to the study’s 34 participants (16 women, 18 men).  Self-report measures of the participants’ emotional reactions were also obtained.  Results indicate that the influence of color appears in the self-reports of emotional experience, but in none of the physiological measures.  These results suggest that people feel, or consciously believe they feel, that color pictures are more pleasing and exciting than monochrome versions of the same images, yet there is no difference in their visceral reactions.  The implications of this dissociation of emotional responses are discussed.


Introduction

Ever since Marshal McLuhan (1964) put forth his provocative aphorism, the medium is the message, media scholars have been interested in the form a message takes as well as its content.  Though McLuhan may have overstated the significance of form or medium characteristics (and understated that of content), he aptly recognized that the influence of media stems, at least in part, from how messages are presented.  The distinction between form and content and its use in analyzing communication existed long before McLuhan (Campbell, 1841), but he deserves credit for recognizing its heuristic value in the examination of (the then new) electronic media.  As media evolve, the concept continues to prove valuable as an entrée into research on the nature of media and their influence.

While a great deal of scholarship has flowed from McLuhan’s seminal idea, research in recent years has gone beyond simply describing the various formal characteristics and labeling different media as either hot or cool, to investigating the impact of message forms.  Using a social scientific approach, researchers have begun to articulate the role that non-content, or formal, attributes of film and television messages play in influencing people’s psychological responses.  For example, studies have shown that picture motion can affect people’s cognitive and emotional responses (Detenber, Simons, & Bennett, 1998; Reeves, Thorson, Rothschild, McDonald, Hirsch, & Goldstein, 1985; Simons, Detenber, Roedema, & Reiss, 1999).  Likewise, research indicates that image size influences message processing and the viewing experience (Detenber & Reeves, 1996; Lombard, 1995; Reeves, Lang, Kim, & Tatar, 1999)1.  In addition to assessing the psychological significance of formal characteristics, these studies have also documented the impact of particular content attributes and some of the ways that form and content interact (e.g., Lombard, Reich, Grabe, Campanella, & Ditton, in press).  In general, form seems to play a lesser role than content in determining psychological responses, yet the influence of these formal attributes is consistent.  It also appears that particular formal characteristics have their greatest impact with certain types of messages.  In other words, it is specific interactions of form and content that influence responses.

As media continue to change, their impact on individuals needs to be assessed.  Investigating the psychological significance of formal properties and their interaction with message content is essential to a complete understanding of how media effects occur.  The present study is an attempt at such an investigation.

Color

The perception of color is essential to our visual experience.  It provides information that helps us understand the physical world and influences how we feel.  Accordingly, a great deal has been written about the nature of color, the perception of it and its aesthetic considerations (Albers, 1987; Birren, 1978; Gombrich, 1960).  In terms of mass media research, color has been the subject of investigations into learning and memory, persuasion, perception and emotional impact.  It has also been studied in the context of both moving and still pictures.  Unfortunately, though the literature on color is extensive, it does not present a uniform set of findings or a consistent perspective on the influence of color.  Indeed, it is the inconclusive nature of research on color that motivates this study.

Early studies on the cognitive effects of color in television presentations indicate that people paid more attention to the details of a television news program when they viewed in black and white (Scanlon, 1967, 1970).  On the other hand, Shaps and Guest (1968) found that more details were recalled from television commercials when they were in color.  Related research on still pictures suggests that in certain contexts color does produce memory advantages (Borges, Stepnowsky, and Holt, 1977; Chute, 1980).  However, a study by Kiphart, Sjogren and Cross (1984) suggests that picture recognition is not influenced by color.  In their study of magazine photographs, Gilbert and Schleuder (1990) also found no advantage for color in terms of recognition accuracy.  Their data did indicate however, that image processing was speeded up by color.  In terms of guiding attention through a newspaper, color has some influence, but mostly it interacts with content and design (Garcia & Stark, 1991).  Results from the same study also showed that color does not influence reading.  In summary, it appears that color in mediated presentations does affect cognitive processing, at least some of the time.

In terms of emotional reactions, the influence of color versus black and white has also been studied in relation to both still and moving images.  A study by Gardner and Cohen (1966) indicates that monochromatic newspaper ads were less attractive to people than the same ads that had some color.  The effect of color seems to depend in part on the content of the picture, though.  Out of 52 images, Winn and Everett (1979) found significant differences in the evaluation of color versus monochrome for only five pictures.  Similarly, a recent study that used physiological measures of emotion in addition to self-reports found some differences between color and black and white for particular still images, but failed to find any overall significant difference (Bradley, Axelrad, Codispoti, Cuthbert, & Lang, 1998).  Participants in the study by Garcia and Stark (1991) however, clearly stated a preference for color in their newspapers.

For moving images, results from social scientific studies on the emotional significance of color are also mixed.  The studies by Scanlon (1967, 1970) suggested that color positively influenced liking of the television presentation.  Similarly, Donohue (1973) found that color increased perceptions of aesthetic quality in political TV advertisements, though women tended to be more influenced then men.  A study by Perse, Pavitt and Burggraf (1991), on the other hand, failed to find any differences due to color in emotional reactions to scenes from feature films.  Furthermore, the differences in responses to colorized films reported by Sherman and Dominick (1988) were only marginally significant.  Even though people say they prefer color television to black and white, participants in a study by Thurman, Ball, Hammack and Walker (1983) actually spent more time watching black and white programming than color.  The only conclusion that can be drawn from these studies is that the influence of color in media presentations is not straightforward.

Film theory has also addressed the role of color, and offers another perspective on its influence.  In general, film theorists contend that color can evoke potent emotional responses in viewers and is therefore one of the director’s most important tools.  For example, Giannetti (1987) says that, “color tends to be a subconscious element in film.  It’s strongly emotional in its appeal, expressive and atmospheric rather than conspicuous or intellectual” (p. 21).  According to Zettl (1990), color serves three functions in film and television: informational, compositional and expressive.  The information color provides can be literal or symbolic.  In either case, it tells us something more about an object or event.  Directors also use color to control and shift emphasis within the frame.  Arranging colors in harmonious and contrasting ways is part of the art of composition.  Lastly, color is used to make viewers feel a specific way (Zettl, 1990).  Though Zettl and others offer many suggestions and guidelines on the use of color for various ends, they admit there is no universal validity for these ‘rules of thumb.’  As Bordwell and Thompson (1989) acknowledge, good or appropriate use of color always depends on the context.

In contrast to color, little has been written about the influence of black and white, or monochrome, film and television presentations on emotional responses.  One notable exception is desaturation theory (Zettl, 1990) which states that for certain types of scenes desaturated colors, or black and white, can produce stronger emotional reactions.  For events that are “extremely internal, such as an intimate love scene, a mother nursing her baby, [or] a wounded soldier waiting helplessly on a battlefield” color may actually dampen viewers’ emotions (p. 79, emphasis in original).  Zettl (1990) suggests that color makes these kinds of internal events external and does not allow people to get involved emotionally.  According to desaturation theory, color leads people to look at rather than into a scene.  Black and white, on the other hand, “invites the audience… to fill in the missing elements,” and typically leads to a more profound emotional experience of the event portrayed (Zettl, 1990, p.80).  While desaturation theory seems quite plausible, it has not been empirically tested.  One of the goals of this study is to see if the theory applies to brief film and television clips.

Dimensions of Emotion

There are two primary theoretical frameworks that exist in the study of emotion.  One conceptualizes emotions discretely or categorically (Izard, 1977; Plutchik, 1980; Ortony, Clore & Collins, 1988) while the other regards emotion as a dimensional construct (Lang, 1995; Osgood, Suci, & Tannenbaum, 1957; Russell, & Mehrabian, 1977).  In general, dimensional approaches posit two or three dimensions that underlie all emotions.  The two most commonly used are autonomic arousal and hedonic valence.  A third, less frequently used dimension, is dominance or control.  Researchers typically describe the dimension of autonomic arousal as a continuous range of affective response that varies from high to low.  That is, at one end of the dimension people feel energized, excited, and aroused and at the other they feel calm, unaroused, or peaceful.  Valence is usually characterized as ranging from pleasant or positive valence at one pole, to unpleasant or negative valence at the other.  These two dimensions, valence and arousal, account for most of the independent variance in emotional responses (Greenwald, Cook, & Lang, 1989), and can be used to effectively discriminate between emotional responses.

One of the benefits of the dimensional view of emotion is that particular physiological measures have specific relationships with the two primary dimensions.  Skin conductance response (SCR) indexes autonomic arousal, while facial muscle movement (i.e., electromyographic, or EMG, activity) and heart rate reveal differences in valence.  In the context of watching moving images, zygomatic activity and heart rate acceleration occurs in response to pleasant stimuli, while unpleasant stimuli are accompanied by corrugator activity and heart-rate slowing (Detenber, Simons, & Bennett, 1998; Lang, 1994; Simons, Detenber & Reiss, in press).  Likewise, as people experience arousal their sweat glands become active and skin conductance responses (SCR) become larger and more frequent (Dawson, Schell, & Filion, 1990; Hopkins & Fletcher, 1994).

Emotion can be measured in different ways.  In the experiment presented here, we focused on subjective and physiological manifestations.  The subjective experience of emotion was assessed using the Self-Assessment Manikin (SAM; Lang, 1980).  This measure, a form of semantic differential scale, requires subjects to introspect and consciously report on their feelings of arousal and hedonic valence.  To assess emotion’s physiological sequelae, we measured facial EMG, skin conductance and heart rate.  Each of these measures yields information about a different aspect of emotional experience, but in general, affective judgments and physiological measures are highly correlated (Greenwald et al., 1989; Lang, Greenwald, Bradley, & Hamm, 1993).  Though the self-report and physiological measures are expected to converge, the physiological measures provide an added benefit – they are less sensitive to any demand characteristics that might be operative during the experiment and, therefore, are less susceptible to impression management.

The present study was designed to explore the relationship between color and people’s emotional responses to film and television clips.  Based on desaturation theory, we expect that the monochrome versions of the images would elicit stronger emotional reactions than the color versions would.  This emotional intensity should appear in greater self-reports of arousal and greater SCRs.  It is also likely that the stronger emotional reactions will manifest themselves in more extreme responses on the valence measures (self-report, facial EMG and heart rate).  That is, the positive images will seem more positive and the negative images will seem more negative when they are seen in black and white.  Lastly, based on prior research, we anticipate that the formal attribute manipulated in the study would interact with the content of the moving images.


Method

Design

This experiment had three within-subject factors: color, clip valence and clip arousal.  Two models were run simultaneously, each with a formal and content variable.  That is, the study used both a 2 (color, black and white) X 3 (positive, neutral and negative valence) and a 2 (color, monochrome) X 3 (low, medium and high arousal) within-subject design.  The categorical emotion variables were created from the self-report ratings of each dimension (see below).  The basic design called for each participant to view 27 film clips in both color and in black & white while physiological measures were taken.  Immediately after viewing each of the 54 clips, subjects rated their emotional reactions.

Subjects

Thirty-five undergraduate students at the University of Delaware received partial credit toward the research participation component of their introductory psychology course or extra credit in their mass communication course.  Potential subjects were disqualified if they reported any form of color blindness.  Data from one of the final 35 participants was eliminated from the study due to technical problems during data collection.  The final sample of 34 consisted of 18 males and 16 females with a mean age of 20.5 years (sd=2.68).

Stimuli

The stimuli consisted of 27 short scenes, or clips, extracted from films and television programs.  The clips chosen for use in the present study were a portion of a much larger set previously standardized by Detenber (1995).  Of the present subset of 27 moving images, 24 were identical to those used by Detenber et al. (1998).  Each clip contained a single shot (no edits) of some scene or object.  The selection of images was based upon ratings obtained from the standardization sample on categories appearing in the International Affective Picture System (IAPS; Lang, Ohman, & Vaitl, 1988).  To facilitate the formation of valence and arousal categories for statistical analysis, the final 27 clips were associated with a wide range of ratings on the emotion dimensions of primary interest in the present study (i.e., valence and arousal).

All stimuli were presented for 6 s and were either monochrome (i.e., black and white; b/w) or color versions of the same clip.  The monochrome version of each clip was created by desaturating the original color version.  That is, using a digital video editing system, the chroma was reduced to zero.  All clips, along with an additional graphic instructing subjects to perform the ratings task, were stored on a video laser disc that was connected to a Macintosh computer.  Stimuli were presented to subjects in one of four orders embedded in a Hypercard program used by the Macintosh to control the sequence and the timing of stimulus presentation.  For two of the four presentation orders,  we created both random clip sequences and a random color-monochrome precedence order (i.e., whether the color or monochrome version was shown first).  The other two sequences had the same clip order, but reversed the color-monochrome precedence to counterbalance the variable.  Although the orders were based on random sequences, they were constrained to insure that the color and monochrome images were never presented on consecutive trials.

Response measurement

Self-report.  Subjects indicated their emotional reactions to each stimulus by providing valence, arousal and dominance ratings in response to each of the 54 moving images on a nine-point paper and pencil version of Lang’s Self-Assessment Manikin (SAM; Lang, 1980).  Using the SAM, valence is rated by marking on or between five graphics depicting the manikin with facial expressions ranging from a broad smile to a severe frown.  Arousal is rated similarly using five graphics depicting the manikin at different levels of visceral agitation, and dominance is rated using manikins which differ in size or prominence in the graphics panel they occupy.  Having subjects mark on or between the graphics allows for more finely tuned ratings and yields a nine-point scale for each dimension.

Physiological recording.  Heart-rate was obtained by attaching a Grass Photoelectric Transducer Model PPS to the subject’s right ear lobe.  The photocell output was fed into a Grass Model 7P1 Low Level DC Preamplifier and Model 7D Driver Amplifier (Bandpass = 1.6 - 3.0 Hz) and then into a series of Coulbourn logic modules for the determination of interbeat intervals.

Skin conductance responses were recorded using a Coulbourn Model S21-22 constant voltage (.5V) skin conductance coupler.  Prior to recording, the palm of the nonpreferred hand was cleansed with distilled water.  Beckman Standard (0.5 cm2) Ag/AgCl electrodes were then placed on the thenar and hypothenar eminence of the palm with Johnson & Johnson KY Jelly used as electrolyte.

Electromyographic (EMG) recordings from the face were obtained by placing Med-Associates miniature Ag/AgCl electrodes over the subject’s left zygomatic and corrugator muscles.  The raw EMG (Bandpass = 3 - 500 Hz) was full-wave rectified and integrated (TC=50 ms) using a Grass Model 7P3 Wideband Amplifier/Integrator.

Procedure

Subjects were provided with a brief description of the stimuli, the ratings task and the recording techniques and then signed an informed consent form.  EMG and skin conductance electrodes were then affixed on their respective recording sites, and the subject was led to an adjacent room equipped with a comfortable arm chair positioned approximately 1.4 m in front of the viewing device (SONY 20" Color Monitor).  The photocell was attached to the ear and the quality of the physiological recordings was inspected.  Subjects then received the complete set of instructions and two 'neutral' practice trials were delivered.  The experiment began if the instructions were understood, if the ratings task was completed properly during the practice trials and if the physiological recordings were free of obvious noise and artifact.

The experiment proper consisted of 54 trials under the control of two laboratory computers -- a 486 PC that initiated each trial and collected the physiological data and the Macintosh that controlled the laser disc player.  At the completion of each 6 s clip, the viewing screen was dark for 1s, and then the instruction to rate their response was presented for 4 s.  Subjects were instructed to rate their emotional response to the image on the three dimensions (valence, arousal, dominance) quickly, and to return their eyes to the viewing screen prior to the appearance of the next image.  The interstimulus interval varied randomly from 17 to 27 s.  Physiological data collection began 2 s prior to the delivery of each image and continued for 10 s.  At the half-way point in the experiment, the experimenter reentered the viewing room to provide a short break for the subject and to ensure that the subject was on the appropriate page in the ratings booklet.  At the conclusion of the experiment, subjects were verbally debriefed and given a brief written explanation of the experiment along with some relevant citations. 

Data Reduction

Our data reduction and analysis procedures were identical to those described in Simons et al., 1999.  In short, skin conductance and facial EMG channels were sampled at 50 cps.  Discrete SCRs that began with an onset latency of 0.5 - 4 s following stimulus onset were identified by visual inspection and quantified as the difference, in uSiemens, between the identified onset point and the peak of the response.

Corrugator and zygomatic data were expressed as the difference between the mean value during the 6 s viewing period and the prestimulus mean.  A derived 'pattern' score was also computed for each trial by first standardizing the data for each muscle within subject and then subtracting the corrugator change from the zygomatic change (Greenwald et al., 1989; Simons et al, in press).  Thus, positive pattern scores would reflect more smiling and less knitting of the brow while negative pattern scores would reflect the opposite. 

Interbeat intervals obtained from the photocell were converted to heart-rate in beats per minute per real-time epoch (500 ms).  When epochs contained portions of two beats, each rate was weighted according to the fraction of the epoch that it occupied (Graham, 1978).  Heart-rate waveforms were the generated by deviating half-second averages during the 7 s post-onset epoch from the half-second average immediately preceding stimulus onset.  Fourteen (7 s) half-second averages, along with the onset point, constituted the heart-rate data that were then submitted to statistical analysis.

Data Analysis

Mean valence and arousal ratings for each of the images were computed collapsing across the monochrome/color dimension.  Valence means were then ranked from most positive to least positive, and then the 27-image set was divided into 9 positive, 9 neutral and 9 negative images.  Likewise, arousal means were ranked from the lowest to highest and the images were divided into 9 low-, medium- and high-arousal categories, again collapsed across the monochrome/color dimension2.  Because the correlation between dominance and valence has exceeded r=0.85 in our previous studies, a separate set of analyses based on dominance categories was deemed redundant with the valence analysis and was not pursued.

Each of the dependent measures was analyzed twice using a repeated-measures Analysis of Variance (ANOVA) with Image Category (Valence or Arousal) and Color as the two within-subject variables.  Single-df orthogonal trends were used to represent the Category variable.  In this analysis, the linear trend (1, 0, -1) is equivalent to the specific contrast of positive v. negative valence or low v. high arousal, whereas the quadratic trend (1, -2, 1) is equivalent to the contrast of the middle category with the two extremes.  The heart-rate analysis also examined single-df orthogonal trends across the half-second data points to assess the effects of both image Category and Color across time.


Results

SAM Ratings

Valence and arousal ratings as a function of both valence and arousal categories are presented in Figure 1.  The two left-hand panels illustrate the impact of the color manipulation on SAM ratings of valence (top) and arousal (bottom).  Clips in color were rated more positively than monochrome clips overall, F(1,33)=8.69, p<.01, and this effect was largest for stimuli in the neutral category, Color X Valence, Fquad(1,33)=17.98, p<.001.  Color clips were also experienced as more arousing than achromatic clips, F(1,33)=8.32, p<.01.  These differences were small in magnitude, but as the figure indicates, they were consistent across subjects.

The two right-hand panels of Figure 1 illustrate the relationship between valence and arousal ratings.  Both positive and negative clips were rated as more arousing than the neutral images, Fquad(1,33)=33.43, p<.001, and negative clips were rated as more arousing than positive clips, Flin(1,33)=8.62, p<.01.  Likewise, high arousal images were rated as somewhat more positive than low or medium arousal images.  Again, the relationship was both linear and quadratic, Fquad(1,33)=19.87, p<.001; Fquad(1,33)=4.84, p<.05. 

Facial EMG

Figure 2 illustrates the relationship between the facial EMG pattern score and both valence and arousal categories.  As expected, positive clips prompted facial activity that was consistent with smiling (i.e., a positive pattern score) and negative clips were characterized by more brow (corrugator) and less smile (zygomatic) activity.  Statistically, the pattern score varied linearly with clip valence, Flin(1,34)=28.25, p<.001, and did not vary significantly with arousal.  This linear relationship between valence and facial EMG pattern was evident in both of the constituent facial muscles, but in this experiment, the effect was statistically significant in the zygomatic, Flin(1,34)=20.86, p<.001, but not quite so for the corrugator, Flin(1,34)=2.75, ns.  Facial expression did not vary with color, F<1, and color did not interact with either image valence or arousal, F’s<1.

Heart Rate

Heart-rate slowed, beginning shortly after stimulus presentation and remained below baseline for the duration of the recording interval.  The half-second by half-second data are presented in Figure 3 as a function of image valence (left-hand panel), image arousal (center panel) and color (right-hand panel).  The trend analysis of variance confirmed the reliability of the deceleration with significant linear (Flin(1,33)=45.09, p<.001) and quadratic (Fquad(1,33)=7.21, p<.01) trends across the presentation period accounting for 85% and 8% of the variance across the half-second data points respectively.

Like the facial EMG, the heart-rate change was significantly related to the valence but not the arousal properties of the stimuli.  As in many previous studies (e.g., Greenwald et a., 1989; Simons et al., 1999) there was a mid-interval acceleratory component that was associated with the positively valenced stimuli.  The relationship between stimulus valence and heart rate was confirmed by the significant linear valence X quadratic half-second interaction, Flin(1,33)=5.30, p<.05.

The impact of the color manipulation on the heart-rate response was weak.  The right-hand panel of Figure 3 suggests that color may prompt a more deceleratory response, but this effect did not approach statistical significance, Flin(1,33)=2.31, p>.10.  And, there was actually poorer discrimination in heart rate among the three valence categories when the clips were presented in color than when the same clips were presented monochromatically.  This effect is illustrated in Figure 4 and supported statistically by a  significant three-way interaction among color, valence and quadratic time Fquad(1,33)=5.45, p<.05.

Skin Conductance

As expected, skin conductance response magnitude reflected the arousal properties of the clips, Flin(1,34)=17.73, p<.001.  As the right-hand panel of Figure 5 illustrates, high-arousal images evoked the largest skin conductance responses.

The left-hand panel of Figure 5 depicts the relationship between SCR magnitude and image valence.  There were no aspects of this relationship that reached statistical significance.  Color did not exert a main effect on the skin conductance response (F<1) nor did it interact with either valence or arousal.


Discussion

The findings of this study do not support predictions based on desaturation theory.  Furthermore, the results provide a somewhat equivocal view of the emotional impact of color in moving images.  The data indicate that color, a formal characteristic or presentation variable, has a small but consistent effect on the subjective experience of emotion, but no appreciable effect on the physiological component of emotional experience.  The differential influence of color on the two sets of measures creates an unusual but not unique pattern of emotional response.

Self-reports versus Physiology

In this study, color affected both the valence and arousal dimensions of emotional experience, but only in affective judgments.  Specifically, when the film clips were seen in color they elicited self-reports of significantly greater arousal and more positive ratings than when they were seen in black and white.  Color also interacted with the content of the clips for the valence dimension, but not for arousal.  Though the absolute difference in the ratings for color and monochrome versions on the two dimensions was small, so was the error variance.  Hence, they achieved significance.

In contrast, color had essentially no impact on the visceral aspect of emotional experience.  That is, the participants’ bodies were not acting as though different emotions were being experienced.  While the data provide some indication that color may influence heart rate, it is difficult say what this might mean.  The separate heart rate waveforms for color and monochrome clips (Figure 4) reveals the three-way interaction between color, valence category and quadratic time.  The data indicate that heart-rate deceleration patterns did not reflect the valence differences in the clips when they were seen in color.  For the monochrome versions of the clips, however, heart rate nicely delineates the valence categories.  This pattern does not match that found in previous research, nor does it fit with existing theory or predictions.  We are somewhat at a loss as to what this difference means or why it occurred.  One might be inclined to suggest that color somehow interfered with or obscured the hedonic nature of the images.  However, this explanation does not comport with the numerous studies that found nicely discriminated heart-rate waveforms for positive, neutral and negative color stimuli.

While color did not have a statistically significant effect on any of the physiological measures, the content of the film clips did.  For the valence-sensitive measure of facial EMG, positive film clips had higher scores (i.e., more smiling and less knitting of the brow) and negative clips had lower scores (more knitting of the brow and less smiling).  The neutral clips had scores in between (Figure 2).  Each valence group of film clips was significantly different from the others.  Similarly, the arousal-sensitive measure of SCR varied in response to the arousal quality of the film clips (Figure 5).  Specifically, the clips that were rated as medium arousal elicited significantly greater SCRs than did the low-arousal clips.  The high-arousal clips elicited significantly greater SCRs than both the medium- and low-arousal clips.  For heart rate, the waveform for the complete set of stimuli clearly reflects the valence of the clips (Figure 3).  The pattern of negative clips leading to greater heart rate deceleration followed by neutral then positive clips is consistent with previous research.  In short, all of these findings are consistent with previous research and substantiate the validity of the physiological measures.

The discrepancy in the data of the different measures of emotion does not necessarily mean that the relationship between color and emotion was spurious or that the study was flawed.  Rather, it could mean one of several things.  First, it is possible that the self-report measures were more sensitive to changes in emotion than the physiological measures, and were able to detect apparently small effects of color.  Even though the error variance was considerably smaller for the SAM measures, it is difficult to make such a comparison, given how different conscious reflection and physiology are.  Alternatively, self-reports tend to be more sensitive to demand characteristics than physiological measures.  There may have been something in the experimental procedure (e.g., seeing both versions of the clips on a single, color monitor) that suggested color should be experienced differently than black and white.  Although we feel this is unlikely, it remains a possibility.

The third explanation for the dissociation found in the results of the study seems to us the most probable.  Emotion manifests itself in three separate systems: the physiological, the psychological (i.e., subjective experience), and the behavioral (i.e., physical actions such as approach and avoidance), and each of these are associated with particular measures (Lang, 1995; Simons et al., in press).  Self-reports and physiological measures tap different systems and index different facets of emotion.  Though they are often correlated, the different types of measures do not necessarily need to be.  In fact, it is not uncommon in emotion research for this kind of dissociation to occur in particular circumstances.  For example, subjects often deny any emotional reaction to same-sex erotica, yet their SCRs indicate that they are indeed experiencing arousal.  Similarly, some phobics report extremely high levels of emotion when confronted with the object of their fears or asked to imagine it, yet their physiological responses do not reflect the terror they experience.  While measures of the three systems of emotion generally converge, situations exist where they do not.  It seems likely the results of this study constitute another instance where one aspect of emotion is affected and another is not.

The Significance of Color

Compared to previous studies of formal attributes the results of this study do not present a compelling case for the influence of color.  However, a few comments about the psychological significance of color can be made.  First, the results of the study do not support the contention of desaturation theory that removal of color from a scene will enhance or intensify emotional reactions.  In fact, it was the color versions of the clips that elicited stronger emotional reactions.  None of the interactions supported the theory either.  That is, the responses to the monochrome clips were not more powerful for any of the content categories, except for the neutral valence images.  This is particularly interesting considering the stimulus set contained examples of the very kind of scenes Zettl (1990) said would benefit most from the absence of color (e.g., a close-up of a girl’s crying face, or soldiers removing a dead body from a battlefield).  In spite of its intuitive appeal desaturation theory still needs empirical support.

Second, based on the results of this study, it appears that color’s influence is only manifest in the conscious evaluation of emotional experience.  That is, people believe or “know” that color is somehow inherently better (i.e., more pleasing and exciting) than black and white and they are able to tell us so in their self-reports.  This belief may be explicitly learned (e.g., the TV salesperson or product brochure said color is better) or influenced by culture or social context (e.g., a social norm that says technological advances are always an improvement).  Whatever the origin of this belief, its impact on the emotional response to color is apparently limited to the conscious experience of emotion.  Despite what our culture or the TV salesperson may say, our visceral experience of emotion is unaffected by color in moving images.

The pattern of color effects found in this study differs from results of previous research on the emotional impact of other formal attributes.  In studies of both motion (Detenber et al., 1998; Simons et al., in press) and screen size (Reeves et al., 1999) significant physiological effects were found for the arousal dimension in addition to significant differences in self-reports of arousal.  This kind of convergence across different measures of emotion contributes to our confidence that certain formal attributes are psychologically meaningful.  However, the absence of convergence in the present study suggests that color is a weak manipulation, at least compared to the influence of image content, motion and screen size.

In addition to comments about the influence of color on emotional responses, a few suggestions can be made for future research on the effects of color in moving images.  Since it is unclear what, if any, impact color has on heart rate, this relationship should be explored further.  In addition, color’s influence on the more mindful aspects of emotional experience as well as its impact on other cognitive processes deserves attention.  To that end, a different set of dependent measures might yet reveal more powerful effects of color.  Thought-listing procedures or other types of self-report could detect some aspects of color’s influence.  Lastly, a different set of stimuli, preferably longer segments, should be used to see if the influence of color changes or diminishes over time.  These modifications to the experimental paradigm, along with replication data would contribute to a better understanding of the significance of color in film and television.


References

Albers, J. (1987).  Interaction of color (revised edition). New Haven, CT: Yale University Press.

Birren, F.  (1978).  Color and human response. New York: Van Nostrand Rheinhold.

Bordwell, D., & Thompson, K.  (1989).  Film art (second edition). New York: McGraw-Hill.

Borges, M. A., Stenowsky, M. A., & Holt, L. H.  (1977).  Recall and recognition of words and pictures by adults and children.  Bulletin of the Psychonomic Society, 9(2), 113-114.

Bradley, M. M., Axelrad, J., M. K., Codispoti, M., Cuthbert, B. N., & Lang, P. J. (1998).  Film noir: Emotion in black in white. Paper presented to the annual conference of the Society for Psychophysiological Research, Denver, CO, USA.

Campbell, G.  (1841).  The philosophy of rhetoric. New York: Harper & Brothers.

Chute, A. G.  (1980).  The effect of color and monochrome versions of a film on incidental and task-relevant learning.  Educational Communication and Technology, 28 (1), 10-18.

Detenber, B. H.  (1995).  The effects of motion and image size on affective responses to and memory for pictures.  Unpublished Doctoral Dissertation, Stanford University, Stanford, CA.

Detenber, B. H., & Reeves, B.  (1996).  A bio-informational theory of emotion:  Motion and image size effects on viewers. Journal of Communication, 46(3),66-84.

Detenber, B. H., Simons, R. F., & Bennett, G. G.  (1998).  Roll ‘em!: The effects of picture motion on emotional responses. Journal of Broadcasting & Electronic Media, 42(1), 113-127.

Donohue, T. R.  (1973).  Viewer perceptions of color and black-and-white paid political advertising.  Journalism Quarterly, 50, 660-665.

Garcia, M. R., & Stark, P.  (1991).  Eyes on the news.  St. Petersburg, FL: The Poynter Institute for Media Studies.

Gardner, B. C., & Cohen, Y. A.  (1966).  ROP color and its effect on newspaper advertising. Journal of Marketing Research, 3, 365-371

Giannetti, L. D.  (1987). Understanding movies (fourth edition). Englewood Cliffs, NJ:  Prentice-Hall.

Gilbert, K., & Schleuder, J.  (1990).  Effects of color and complexity in still photographs on mental effort and memory. Journalism Quarterly, 67, 749-756.

Gombrich, E. H.  (1960).  Art and illusion: A study in the psychology of pictorial perception. Princeton, NJ: Princeton University Press.

Grabe, M. E., Lombard, M., Reich, R., Bracken, C. C., & Ditton, T. B. (1999).  The role of screen size in viewer experiences of media content. Visual Communication Quarterly, 6(2), 4-9.

Greenwald, M.K., Cook, E.W. & Lang, P.J.  (1989).  Affective judgment and psychophysiological response: Dimensional covariation in the evaluation of pictorial stimuli.  Journal of Psychophysiology, 3, 51-64.

Hopkins, R. & Fletcher, J. E.  (1994).  Electrodermal measurement: Particularly effective for forecasting message influence on sales appeal.  In A. Lang (Ed.),  Measuring psychological responses to media (pp. 113-132).  Hillsdale, NJ: Lawrence Erlbaum Associates.

Izard, C. E.  (1977).  Human emotions.  New York: Plenum.

Kiphart, M. J., Sjogren, D. D., & Cross, H. A.  (1984).  Some factors involved in complex-picture recognition.  Bulletin of the Psychonomic Society,22(3), 197-199.

Lang, A.  (1994).  What can the heart tell us about thinking?  In A. Lang (Ed.),  Measuring psychological responses to media. (pp. 65-84).  Hillsdale, NJ: Lawrence Erlbaum Associates.

Lang, P. J.  (1980).  Behavioral treatment and bio-behavioral treatment:  Computer applications.  In J. B. Sidowski, J. H. Johnson & T. A. Williams (Eds.), Technology in mental health care delivery systems (pp. 119-137).  Norwood, NJ:  Ablex.

Lang, P. J.  (1995).  The network model of emotion:  Motivational connections.  In R. S. Wyer & T. K. Srull (Eds.),  Advances in social cognition (Vol. 6).  Hillsdale, NJ:  Lawrence Erlbaum Associates.

Lang, P. J., Greenwald, M. K., Bradley, M. M., & Hamm, A. O.  (1993).  Looking at pictures:  Affective, facial , visceral, and behavioral reactions. Psychophysiology, 30, 261-273.

Lang, P. J., Öhman, A., & Vaitl, D.  (1988).  The international affective picture system [Photographic slides].  Gainesville: University of Florida, The Center for Research in Psychophysiology.

Lombard, M.  (1995).  Direct responses to people on the screen:  Television and personal space. Communication Research, 22, 288-324.

Lombard, M., Reich, R. D., Grabe, M. E., Campanell, C. M., & Ditton, T. B. (in press).  The impact of television: Screen size and viewer responses to point-of-view movement. Human Communication Research.

McLuhan, M.  (1964).  Understanding media: The extensions of man. New York: Penguin.

Ortony, A., Clore, G. L., & Collins, A.  (1988).  The cognitive structure of emotions.  Cambridge: Cambridge University Press.

Osgood, C., Suci, G., & Tannenbaum, P.  (1957).  The measurement of meaning.  Urbana, IL: University of Illinois Press.

Perse, E. M., Pavitt, C. Q., & Burggraf, C. S.  (1991).  Effects of color and black-and-white video on activation of pro- and antisocial schemata.  Paper presented at the Speech Communication Association convention, Atlanta, GA, USA.

Plutchik, R. (1980).  Emotions: A psychoevolutionary synthesis.  New York: Harper &Row.

Reeves, B., Lang, A., Kim, E.Y. & Tatar, D.  (1999).  The effects of screen size and message content on attention and arousal. Media Psychology, 1, 49-67.

Reeves, B. E., Thorson, E., Rothschild, M., McDonald, D., Hirsch, J., & Goldstein, R.  (1985).  Attention to television: Intrastimulus effects of movement and scene changes on alpha variation over time. International Journal of Neuroscience, 25, 241-255.

Russell, J., & Mehrabian, A.  (1977).  Evidence for a three-factor theory of emotions. Journal of Research in Personality, 11, 273-294.

Scanlon, T. J.  (1967).  Color television: A new language?  Journalism Quarterly, 44, 225-230.

Scanlon, T. J.  (1970).  Viewer perceptions of color, black and white TV: An experiment.  Journalism Quarterly, 47, 366-368.

Schaps, E., & Guest, L.  (1968).  Some pros and cons of color TV.  Journal of Advertising Research, 8,(2) 28-39.

Sherman, B. L., & Dominick, J. R.  (1988).  Perceptions of colorization.  Journalism Quarterly, 65, 976-980.

Simons, R. F., Detenber, B. H., Roedema, T. M., & Reiss, J. E. (1999).  Emotion-processing in three systems: The medium and the message. Psychphysiology, 36, 619-627.

Thurman, J. G., Ball, S. E., Hammack, B., & Walker, S. J.  (1983).  “Free operant” television viewing as a function of color, sound, and time of day.  Perceptual and Motor Skills, 57, 1043-1047.

Winn, W., & Everett, R. J.  (1979).  Affective rating of black-and-white pictures. Educational Communication and Technology, 27 (2), 148-156.

Zettl, H.  (1990).  Sight sound motion; Applied media aesthetics. ( 2nd Ed.), Belmont, CA: Wadsworth Publishing.

Footnotes

1.      For a recent review of the psychological significance of screen size see Grabe, Lombard, Reich, Bracken, & Ditton (1999).

2.      The range of SAM valence rating for each category were: Positive (5.91-7.78),  neutral (4.51-5.87) and negative (2.40-4.50).  The range of SAM arousal means for the three arousal categories were: low (2.38-3.43), medium (3.44-4.16) and high (4.47-5.79).

 

Return to:

Simons Homepage