The Journal of General Psychology, 2006, 133(2), 191-207 Copyright © 2006 Heldref Publications
Affective Impressions and Memorability of Color-Form Combinations YUIKO SAKUTA JIRO GYOBA Department of Psychology Tohoku University
ABSTRACT. In this study the authors investigate how various impressions affect the recognition of color-form pairs. They performed a preliminary study, a main study of impression rating using the semantic differential method, and a recognition experiment. On the basis of the preliminary study, the authors chose 14 appropriate pairs of adjectives and 48 color and form stimuli and used them for the main study. Factor analysis of the main study extracted three factors: Activity, Potency, and Evaluation. In the recognition test, the authors found that color-form pairs with congruent impressions in Activity or Potency were more easily recognized than pairs with incongruent impressions. In contrast, pairs with incongruent impressions in Evaluation were recognized more easily than congruent pairs. These results are discussed in relation to the assumed network representations of affective impressions that have different characteristics depending on the three factors. Key words: color and form, impression, recognition memory, semantic differential method
HUMANS RECOGNIZE OR MEMORIZE OBJECTS in their daily lives that consist of combinations of many attributes, such as several parts, colors, or textures. We are interested in trying to determine what types of combinations tend to be more memorable. Many researchers have shown that congruent information with schema or context can be recognized well (Bellezza & Bower, 1981; Cohen, 1977; Cohen, 1981; Judd & Kulik, 1980; Smith & Graesser, 1981), and some researchers have indicated that incongruence in context or expectance encourages memorability (Hamilton & Gifford, 1976; Hastie & Kumar, 1979; Srull, 1981; Whittlesea & Williams, 2001). This research was funded by a Japanese Society for the Promotion of Science Grant-in-Aid for Scientific Research (No. 16650043), sponsoring Jiro Gyoba. This manuscript was edited by Ms. K. Miller, Royal English Language Centre, Fukuoka, Japan. Address correspondence to Yuiko Sakuta, Department of Psychology, Graduate School of Arts & Letters, Tohoku University, Kawauchi27-1, Aoba-ku, Sendai, 980-8576, Japan ;
[email protected] (e-mail).
Recently, many researchers (e.g., Bower, 1981) focusing on affective aspects in cognitive processes have examined the affective value of stimuli or the emotional states of participants. In studies of mood congruency effects (Bower, 1981; Isen, Shalker, Clark, & Karp, 1978) or negativity bias (Hamilton & Zanna, 1972; Kanouse & Hanson, 1972; Pratto & John, 1991), researchers have believed that humans can preferentially process important information or systematically store such information so that it may be accurately remembered. In this study, we investigated how various combinations of the various impressions produced from stimuli affect recognition or memorability. We defined an impression as a “highly subjective image that a person affectively receives from a stimulus.” For example, a person not only feels “this painting is good (or bad),” but also “this painting is warm” or “this painting is strong.” Such impressions are a type of subjective or holistic information and reflect particular affective feelings or responses. It is important to understand how combinations of various impressions affect recognition because people are faced with various combinations of impressions in daily life. Researchers have investigated the characteristics of the impressions systematically and quantitatively using the Semantic Differential (SD) method (Osgood, Suci, & Tennenbaum, 1957). Semantic Differential Method In the 1950s, Osgood et al. (1957) developed the SD method to measure the affective meanings of language on the basis of the hypothesis that words include both literal and affective meanings. With the SD method, various bipolar adjectives are used to judge multiple impressions received from the materials. Then, the multiple impressions are described within fewer dimensions based on correlations among the ratings of the adjectives used. Osgood et al. suggested that SD data can be summarized using three main factors: Activity, Potency, and Evaluation. Activity is represented by adjectives such as active–passive or cheerful–cheerless. Potency is represented by adjectives such as powerful–weak or hard–soft, and Evaluation is represented by adjectives such as beautiful–ugly or likable–unlikable. These factors have been confirmed among many materials (concepts or objects), cultures, and countries (Osgood et al., 1957; Osgood, 1960). Since its creation, the SD method has been used in various impressions studies. However, there have been relatively few studies in which researchers examine the effect of these factors on the memorability of stimuli. For example, Winograd (1966) indicated that words that were close to each other on the semantic differential factor dimension were easily confused. Turvey and Fertig (1970) discussed the effect of the polarity of factors on proactive interference and indicated that words on the same end of a dimension resemble each other in terms of affective meaning compared with words on the opposite end of a dimension, or on a different dimension. Wickens and Clark (1968) and Wickens (1970) reported that words categorized on different factor dimensions or with opposite polarities on the same dimension are encoded and stored differently than words categorized on the same factor dimension with the same polarity. The researchers thought that word stimuli are stored using affective categories based on the SD dimensions. Winograd’s, Turvey and Furtig’s, and Wickens and Clark’s results indicate that Osgood’s (1957) three factors substantially affect the memory process. Recently, Xiong, Franks, and Logan (2003) also used the SD factors as learning categories in a priming experiment. The priming effect was largest when the words were in the same category, whereas there was little priming effect when
there was a large distance in the factor space of the words. Therefore, it seems that words in the same factor category are recognized based on a common cognitive process. Conversely, several researchers (Osgood et al., 1957; Oyama, 2003; Oyama, Yamada, & Iwasawa, 1998) have examined combinations of stimuli using the SD method. Of particular interest are studies in which researchers have investigated how individual impressions of colors or shapes contribute to predicting the overall impression of a combined stimulus containing a color and shape or a combination of two colors (Oyama, 2003; Oyama, Yamada, & Iwasawa, 1998). Osgood et al. (1957) found a characteristic effect specific to the Evaluation factor, called pessimistic evaluative stickiness. They found that overall impressions of combined stimuli could not only be accurately predicted from the individual stimuli, but also the overall impressions tended toward the negative direction when positive words (high Evaluation) were combined with negative words (low Evaluation). Hogg (1969) noted differences between the predicted and observed values regarding scales of pleasant–unpleasant and strong–weak. The rated values for the color pairs were not much more pleasant or strong than were the values predicted from regression formula. Oyama et al. (1998) and Oyama (2003) suggested that Activity and Potency had additive properties whereas Evaluation had a nonadditive property. This conclusion was based on multiple regression analyses of the factor scores of combined stimuli that consisted of colors and forms (or video images and music) and the factor scores of colors and forms (or video images and music) that were independently presented. The Evaluation factor tends to reveal a complex interaction, in which the impression of combined stimuli is likely to shift in a negative direction and does not correspond to the values predicted from the impressions of individual stimuli. In contrast, the Activity and Potency factor scores of combined stimuli can be predicted from the factor scores of individual stimuli. A major focus of our study was to investigate the reasons why these different characteristics are observed for the three factors. Positivity and Negativity Many researchers (Hamilton & Zanna, 1972; Kanouse & Hanson, 1972; Pratto & John, 1991) have compared memory traits and cognitive properties on positive and negative stimuli, finding support for an effect called negativity bias, in which negative stimuli draw attention strongly and are therefore more memorable than positive stimuli. Moreover, some researchers (Taylor & Brown, 1988; Taylor, 1991) have proposed the mobilization–minimization hypothesis, which states that humans try to minimize the impact of negative events as time passes. For example, enhanced sensitivity to negative stimuli would be adaptive when one must quickly identify harmful stimuli in the environment. However, it is not good for people’s mental health if the negative state lasts for a long time. Several researches (Thurstone, 1931; Cacioppo & Berntson, 1994) have measured positivity and negativity in the context of attitude studies. Thurstone assumed that the attitudes of like and dislike were the opposite ends of a single bipolar continuum. This hypothesis is broadly supported and such positive and negative processes underlying attitudes have traditionally been measured on bipolar scales in questionnaires. Cacioppo and Berntson, however, advocate an evaluative space model, which supposes that attitudes do not always have such bipolarity. According to their model, positivity and negativity are not on opposite ends of a single spectrum, but are represented within a bivariate space. In respect to Osgood’s (1957) three factors, Activity and Potency mainly contain adjective
pairs such as bright–dark or hard–soft, which are not associated with clear positive or negative affects. These two factors are based on information related to sensory modalities. In contrast, Evaluation contains many adjective pairs, such as likable–dislikable and beautiful–ugly, in which one adjective has a positive meaning and the other has a negative meaning. Thus, positivity and negativity are highly related to the Evaluation factor in Osgood’s three factors. It is likely that these fundamental properties cause the difference between the additive properties of Activity and Potency and the nonadditive property of Evaluation in predicting factor scores. Similarity of Affective Meanings and Semantic Network Theory Using photographs of faces and name cards, we have previously investigated how combinations of impressions affect the recognition of face and name card pairs having either congruent or incongruent impressions in the three SD factors (Sakuta & Gyoba, 2003). We found that pairs with congruent impressions in the Activity or Potency factor were recognized more easily than were those with incongruent impressions. In contrast, pairs with incongruent impressions in the Evaluation factor were recognized more easily than were pairs with congruent impressions. We tried to interpret the results we obtained in our 2003 study for the face and name card pairs based on the semantic network activation hypothesis (Collins & Loftus, 1975). According to this hypothesis, concepts with similar meanings are located close to each other in the semantic network. A link between concepts might be stimulated when particular information about one concept is entered, and such concepts may then be automatically activated by each other. We have expanded this hypothesis to include affective meanings or impressions and have assumed that concepts with similar impressions might be represented in close proximity in the network of affective impressions. As described above, Winograd (1966) and Turvey and Fertig (1970) have indicated that words located close to each other in factor space were easily confused in a study measuring the affective meanings of words using the SD method. We have previously reported that pairs with similar impressions were recognized accurately in Activity and Potency (Sakuta & Gyoba, 2001, 2003). Consequently, it is likely that two concepts with similar impressions are located more closely within a network representing affective meanings. If so, these concepts may activate each other and may be effective cues for triggering memory of one concept when the other is presented as a stimulus. However, in Evaluation, pairs with incongruent impressions are more easily recognized than pairs with congruent impressions. Therefore, it is possible that congruent information in Evaluation is not represented closely. Because Evaluation involves a nonadditive property (Osgood et al., 1957; Oyama, et al., 1998; Oyama, 2003), it is possible that positivity and negativity of the stimuli are not represented linearly on the network of affective impressions, but are mapped nonlinearly. In other words, likable and dislikable are not necessarily located far from each other within the network. Consequently, a different interaction would be induced when the paired stimuli have congruent or incongruent impressions in Evaluation. Purposes of this study The faces and name cards that we used in our 2001 and 2003 studies are complex stimuli that contain numerous elements; thus, there may be alternative interpretations of these data. Therefore, we decided to replicate our previous study (Sakuta & Gyoba, 2003) using simplified
stimuli, colors and forms, to investigate whether the effects found in our previous study can be generalized. There are two purposes of our study: (a) to examine whether the unique interaction of Evaluation can be seen using colors and forms as well as faces and name cards, and (b) to clarify how the combined impressions of colors and forms differently affect recognition depending on three factors addressed by Osgood et al. (1957). We hypothesized that colors and forms with similar impressions in Activity or Potency are represented closely within the network and can facilitate recognition of each other. In contrast, colors and forms with similar impressions in Evaluation may not be mapped closely within the network, resulting in lower recognition of congruent pairs. If such results were found to indicate a general tendency of recognition, we may have discovered a clue that will help us elucidate the processes by which humans recognize or memorize multiple stimuli containing various affective impressions. IMPRESSION RATINGS OF COLORS AND FORMS Preliminary Study We created color and form stimuli and measured the properties of their impressions using the SD method. First, we performed a preliminary study to select appropriate stimuli and adjectives. Method Participants. The participants were a total of 20 male and female undergraduates and graduate students at Tohoku University who were given class credit for participation. All of the students were native Japanese speakers. Stimuli. We generated 60 geometric forms by superimposing an oscillatory closed contour between two virtual circles using Adobe Photoshop. Four variables were manipulated to produce many variations: (a) the property of lines used to generate contours: sinusoidal or straight; (b) the frequency of contour oscillations: 4, 5, 7, 10, or 20; (c) the amplitude of contour oscillations: shallow, medium depth, or deep; and (d), the regularity of contour oscillations: regular or irregular. We also added 12 stimuli that were generated by 45 rotations of the forms with contour frequencies of 4 or 5 because there was a possibility that different impressions might be produced by these rotatory transformations. Using the standard color chips of the Munsell color system, we chose 60 colors as color stimuli. From every 10 main hues and several other hues, we selected 4 or 5 colors. These colors were reproduced and presented on a computer display. Three experimenters (authors and one undergraduate student) subjectively matched the colors on the display with the color chips. They adjusted the colors on the display until more than one of the three experimenters agreed. They randomly divided the colors and forms into two groups and each of the participants rated the stimuli contained in one of the groups. Moreover, we prepared two different orders of stimuli presentation for each group to reduce any possible effects of the order of presentation. Procedure. The stimuli were presented one by one against a black background on a 15-inch monitor. For the color stimuli, a square form (10 10 cm) was presented and filled with one of the colors. For the form stimuli, a square of the same size was presented in homogeneous white. The experiment was conducted in a room illuminated by four D65 fluorescent lights, which were
produced by Toshiba Lighting and Technology Corporation, Tokyo, Japan. They are commonly recommended for color tests and comparisons and satisfied the criteria of JIS (JIS Z 8723). Their color temperature was 6500K and their color rendering properties were AAA. The participants rated the presented colors and forms one by one using the SD method in Japanese. On the basis of previous studies (Nakano, 1972; Chijiiwa, 2001; Inoue & Kobayashi, 1985) and our previous research (Sakuta & Gyoba, 2003), we used 23 pairs of adjectives that were applicable for rating both color and form. We prepared two types of questionnaire sheets in order to counterbalance any possible effect of the order and polarity of the adjectives. It took participants about 40 to 50 m to complete the rating task. Results We first analyzed the color and form ratings separately and found very similar factor structures between the two sets of data. Then, we performed further factor analysis on the combined data and extracted three factors using the principal factor method with varimax rotation. Consequently, the three extracted factors could be regarded as Potency, Activity, and Evaluation. We then eliminated several unsuitable adjectives and selected 14 adjective pairs for the main study on the basis of the following rules: We eliminated adjectives that (a) did not belong to any factor, (b) had high factor loading on several factors, (c) had low communality. Table 1 shows the 14 selected adjective pairs. We used the Japanese words that are indicated under the English words in the table. Next, we selected appropriate stimuli for the main study. When two stimuli had very similar rating scores or factor scores, we chose one of the stimuli and excluded the other. Finally, we chose 50 color stimuli and 50 form stimuli (see Appendix). Main study On the basis of the preliminary study, we performed the main study of impression ratings using different participants. Method Participants. The participants comprised 48 undergraduate and graduate students who had not participated in the preliminary study. All of the students were native Japanese speakers. Stimuli. We divided the previously chosen 50 colors and 50 forms into two groups and prepared two different orders of stimuli presentation for each group to reduce any possible effects of the order of presentation. Procedure. The procedure was almost identical to the procedure we used in the preliminary study. We used 14 pairs of adjectives that had been selected in the preliminary study. The task took participants approximately 15 to 30 m to complete. Each half of the participants rated half of the colors and half of the forms on 7-point scales. The rating order was counterbalanced among the participants. Results
We performed factor analyses separately for the color and form data. We obtained very similar factor structures for both sets of data. Therefore, we performed further analysis on the combined data of both stimuli. We extracted three factors using the principal factor method with varimax rotation, and named these Activity (contribution ratio: 27.35%), Potency (23.03%), and Evaluation (17.32%). Table 1 shows the results of factor loadings. Representative adjectives (all Table 1. Factor loadings after varimax rotation
Factor
Adjectives plain-showy (jimina-hadena)
passive-active (fukappatsuna-kappatsuna)
Activity
static-dynamic (seitekina-doutekina)
quiet-noisy (shizukana-sawagashii)
dark-bright (kurai-akarui)
square-round (kakubatta-marui)
hard-soft (katai-yawarakai)
Potency
tense-lax (kinchoushita-yurunda)
cold-warm (tsumetai-atatakai)
masculine-feminine (danseitekina-joseitekina)
beautiful-ugly (utsukushii-minikui)
likable-unlikable Evaluation
(sukina-kiraina)
elegant-inelegant (hin'noaru-hin'nonai)
organized-disorganized (matomatta-barabarana)
Eigen value Contribution of each factor Cumulative contribution
Activity Potency Evaluation 0.882
-0.008
-0.094
0.879
0.007
-0.028
0.835
0.041
0.152
0.822
-0.050
0.260
0.727
0.305
-0.293
-0.102
0.880
-0.042
-0.045
0.879
-0.059
-0.079
0.801
0.081
0.241
0.734
-0.122
0.152
0.614
-0.188
-0.190
-0.072
0.866
-0.105
-0.107
0.797
0.242
-0.030
0.726
0.413
-0.071
0.516
3.828 27.345 27.345
3.224 23.032 50.377
2.425 17.323 67.699
Note: All adjectives were presented in Japanese as indicated under the English words.
of them were presented in Japanese) for each factor were as follows: plain–showy (jimina–hadena), passive–active (fukappatsuna–kappatsuna), and static–dynamic (seitekina–doutekina) for the Activity Factor; hard–soft (katai–yawarakai), square–round (kakubatta–marui), and masculine–feminine (danseitekina–joseitekina) for the Potency Factor; beautiful–ugly (utsukushii–minikui), likable-dislikable (sukina–kiraina), and elegant–inelegant (hin’noaru–hin’nonai) for the Evaluation Factor. Next, we chose 24 colors and forms, which we used as target stimuli in the recognition tests based on their properties in regard to Osgood’s (1957) three factors. The stimuli were located on all sides of the axes of the three factors. Then we prepared four types of stimuli pairs: (a) one color and one form, both of which were on the plus side of each factor axis, (b) one color and one form, both of which were on the minus side, (c) one color on the plus side and one form on the minus side, and (d) one color on the minus side and one form on the plus side. The stimuli on the plus side had positive meanings for the factor; for example, plus on Activity means that the stimulus has dynamic and active impressions. Other stimuli, which were located around the center of the factor space and which had weak factor properties, were used as distracter stimuli in the recognition test. RECOGNITION TEST For the impression ratings, we chose color and form stimuli that were located on the periphery of each factor axis, and generated four types of pairs, some with congruent impressions and some with incongruent impressions. We determined whether pairs were more recognizable depending on whether the combination of the impressions of color and form was congruent or incongruent. We regarded the total of correct responses as the recognition accuracy. Moreover, we asked the participants to rate the pairs for congruent or incongruent impressions to confirm whether or not the treatment of congruence of impression had been successful. It is important to consider the effect of the combination of the impression of objects on recognition because many of the objects that surround humans in their daily lives exist in combination with other objects. Method Participants The participants comprised 20 undergraduate and graduate students at Tohoku University (10 men and 10 women) who had not taken part in the impression rating of colors and forms. All of the students were native Japanese speakers. We paid these participants for their participation. Stimuli We used the previously selected 24 target colors and forms, and 24 distracter colors and forms. We prepared two types of combination of the colors and forms for each of the target pairs to reduce the effect of the stimuli themselves, instead of the effect of the congruence or incongruence of their impressions. We assigned male and female participants in equal numbers to the task using each type of combination. The stimuli were presented on a 15-inch computer display with a pixel resolution of 1,024 768. The size of the available viewing area was 327 243 mm.
Procedure In the study phase, we asked participants to memorize 48 pairs (24 target pairs mixed with 24 distracter pairs), which were presented one by one on the computer display. The presentation time for each pair was 4 sec and a slide was inserted to provide a 1 sec ISI (Inter Stimulus Interval). We performed the recognition test immediately after all the pairs had been presented. In the test phase, we presented one form as a recognition cue and three colors, including one target and two distracters, with the form. We asked participants to remember the pairs and choose the color that had been paired with each form during the study phase. We alternately repeated study and test phases five times during each of five sessions. We changed the presentation order of the pairs of stimuli in every session. We computed the total of the correct responses for each pair as the recognition performance. At the end of the experiment, we presented the color-form pairs again and asked participants to rate how the impressions of color and form matched each other on a 7-point scale (fitness rating).
Recognition Performance（％）
Results Figure 1 shows the recognition performance for the pairs of each combination of factor properties. Three-way analysis of variance (ANOVA), factor type (3) color properties (2) form properties (2), demonstrated that the main effects of factor type F(2, 38) = 4.992, p < .05; color properties F(1, 19) = 8.631, p < .001; and form properties, F(1, 19) = 7.395, p < .05, were significant. The interactions between the factor type and color properties, F(2, 38) = 9.156, p < .001; the color properties and form properties, F(1,19) = 14.687, p < .005; and the factor type,
80 60 40 20
Form+ Form－
0 Color Color ＋ －
Color Color ＋ －
Color Color ＋ －
Activity
Potency
Evaluation
Figure 1. Recognition performance for pairs comprising one color and one form, in relation to the three impression factors and the polarity
the color properties, and the form properties, F(2, 38) = 18.809, p < .001, were also significant. The test of simple interaction on the interaction among factor, color, and form revealed significant color-and-form interactions for each of the factors of Activity, Potency, and Evaluation, F(1, 57) = 33.999, p < .001; F(1, 57) = 11.298, p < .005; F(1, 57) = 7.157, p < .01, respectively. For Activity, the recognition performance for pairs of plus color with plus form was significantly higher than that for pairs of plus color with minus form or minus color with plus form F(1, 114) = 25.556, p < .001; F(1, 114) = 25.799, p < .001. Moreover, the recognition performance for pairs of minus color with minus form was significantly higher than that for pairs of plus color with minus form or pairs of minus color with plus form, F(1, 114) = 5.591, p < .05; F(1, 114) = 6.167, p < .05. The pattern of recognition performance for Potency was similar to that for Activity. Pairs of plus color with plus form and minus color with minus form were significantly easier to recognize than pairs of plus color with minus form, F(1, 114) = 15.228, p < .001; F(1, 114) = 32.402, p < .001. In contrast, for Evaluation the recognition performance for pairs of minus color with plus form was significantly higher than that for pairs of plus color with plus form or pairs of minus color with minus form, F(1, 114) = 8.865, p < .005; F(1, 114) = 3.807, p < .10. As described above, a different pattern of interactions was found depending on the type of impression factor. This tendency can be summarized as follows: pairs with congruent impressions on Activity or Potency were easily recognized, whereas pairs with incongruent impressions on Evaluation were easily recognized. Next, we conducted the same analysis on the fitness rating data (Figure 2). Three-way ANOVA demonstrated that the main effects of factor, F(2, 38) = 3.607, p < .05, and the interaction between factor and color, F(2, 38) = 15.666, p < .001; factor and form F(2, 38) = 9.608, p < .001; color and form, F(1, 19) = 37.425, p < .001; and factor, color, and form, F(2, 38) = 17.078, p < .001, were significant. The test of simple interaction revealed significant color-form interactions for both Activity, F(1, 57) = 25.738, p < .001, and Potency, F(1, 57) = 42.141, p < .001. For Activity and Potency, most of the fitness ratings for the plus-plus and minus-minus pairs were significantly higher than those for the pairs with opposite polarities, F(1, 114) = 53.416, p < .001; F(1, 114) = 28.836, p < .001; F(1, 114) = 3.486, p < .10; F(1, 114) = 15.576, p < .001; F(1, 114) = 22.568, p < .001; F(1, 114) = 4.094, p < .05; and F(1, 114) = 52.385, p < .001. For Evaluation, there were no significant differences among the pairs. In summary, the fitness rating was also differently affected by the type of impression factors. For both Activity and Potency, the fitness of pairs with congruent impressions was rated higher than the fitness of pairs with incongruent impressions. In contrast, pairs with incongruent impressions were rated higher than pairs with congruent impressions for Evaluation (n.s.). GENERAL DISCUSSION In the first part of this study, we analyzed the impressions of color and form. Factor analysis revealed a similar factor structure between the ratings of color and form. The extracted factors can be regarded as Activity, Potency, and Evaluation, which correspond to the three factors formulated by Osgood et al. (1957). The recognition experiment demonstrated that pairs which had congruent impressions on Activity and Potency were more easily recognized. In contrast, pairs that had incongruent impressions on
Evaluation were more easily recognized. These results are in agreement with previous experiments using photographs of faces and name cards (Sakuta & Gyoba, 2001, 2003). The results of the fitness ratings are also in agreement with our previous research. Pairs with the same polarity on Activity and Potency revealed higher fitness ratings than pairs with opposite polarity, whereas pairs with the same and opposite polarities on Evaluation did not reveal significant differences. Thus, both this report and our previous studies indicate that impression factors have different effects on memorability and fitness rating. Furthermore, these different characteristics among the factors are not specific to pairs comprising a face and a name card, but can be regarded as general tendencies. One conclusion we have drawn from these results is that the stimuli are categorized according to the impression factors and are processed differently. In our previous studies (Sakuta & Gyoba, 2001, 2003), we discussed these results on the basis of the spreading activation theory (Collins & Loftus, 1975). This theory assumes that similar concepts activate each other and facilitate recognition within the semantic network. This interpretation could be applied to the network of affective meanings or impressions. Spreading activation on the assumed network of affective impressions can be directly applied to the results for Activity and Potency because pairs with similar impressions were more memorable than pairs with different impressions on these two factors. However, the results relating to Evaluation cannot be easily explained by spreading activation theory. Therefore, a more plausible explanation or additional factors are required, especially in terms of the Evaluation factor. One possibility is that there are differences in the network representations. As noted in the introduction, there may be different mapping mechanisms for Activity or Potency and Evaluation. Concepts with similar impressions in Activity or Potency are assumed to be located close to each
Fitness Ratingsof
7 6 5 4 3 Form+ Form－
2 1 Color Color ＋ －
Color Color ＋ －
Color Color ＋ －
Activity
Potency
Evaluation
Figure 2. Fitness ratings for pairs comprising one color and form in relation to the three impression factors and the polarity
other within the affective network representation, whereas concepts with similar impressions in Evaluation are not necessarily located in close proximity. It is likely that such differences can induce different recognition characteristics. Osgood et al. (1957) found a characteristic effect specific to the Evaluation factor, called pessimistic evaluative stickiness. The Evaluation factor tends to reveal complex interactions, in which the impression of combined stimuli is likely to shift in a negative direction and does not correspond to the values predicted from the impressions of the individual stimuli. In contrast, the Activity- and Potency-factor scores of combined stimuli can be predicted from the factor scores of individual stimuli. More recently, Oyama et al. (1998) obtained similar results based on multiple regression analyses of the factor scores of combined stimuli that consisted of colors and forms (or video images and music) and the scores of colors and forms (or video images and music) when they were independently presented. It is possible that such nonlinear interactions observed in the Evaluation factor might affect recognition performance. However, it remains to be determined why these interactions have different effects on the recognition performance of pairs of stimuli on Activity or Potency and Evaluation. James (1969) emphasized the assertion of Osgood (1960) that the affective reaction system, which is thought to generate responses to adjectives in the SD method, is the psychological basis of synesthesia and metaphor. The scales used in the SD method contain adjectives related to various modalities. For example, visual stimuli are often rated by adjectives related to other modalities, such as sweet or hot. The Activity and Potency factors are rated in close association with sensory modalities, (e.g. hard color–soft color, dark form–light form. It is therefore likely that the raters internally produce synesthetic representations, especially during the rating task for Activity and Potency. If this is the case, it would be easy to judge the impression fitness of color-form pairs because impression processing is based upon sensory information or physical features contained by the stimuli, such as hard or soft and dark or light. In fact, we found that it was easy to rate the fitness and reflect the polarities of the pairs on the Activity and Potency factors. The degree of fitness can be assumed to directly affect recognition performance because similar internal representations of Potency or Activity impressions based on sensory processing are expected to activate each other and facilitate recognition performance. Conversely, the rating of Evaluation, which includes adjective pairs such as beautiful–ugly and likable–unlikable, does not have a direct relationship to physical features or sensory information, but is more closely related to affective information. Consequently, Evaluation would be related to subjective, self-relevant, or affective judgments, which are not dependent upon sensory modalities. Therefore, rating the impression fitness of the pairs in Evaluation is expected to become intricate and difficult, as the raters are not able to rely on physical features or sensory information. In fact, the fitness ratings for the pairs on Evaluation did not reveal any significant differences. Furthermore, the involvement of other properties may also be important. Pairs with incongruent impressions on Evaluation (e.g., color is likable whereas form is unlikable) can be assumed to evoke negativity or unexpectedness, both of which are thought to be attention-capturing (Olson & James, 2002). Some studies (e.g., O'Brien & Myers, 1985) indicated that inconsistent information is deeply elaborated during processing and that such elaboration elicits distinctiveness of the information in the memory. If this is the case, it is likely that the attention-capturing properties of negativity or unexpectedness and the distinctiveness produced by elaboration, enhance the memorability of pairs with incongruent impressions on Evaluation.
In contrast, we can assume that pairs with incongruent impressions on Activity or Potency (e.g., active color and passive form) do not produce such large negativity or unexpectedness, since the elaboration process may not play a central role in judgments of Activity or Potency. Both incongruent and congruent pairs may be processed on sensory or physical information, and similar internal representations of Potency or Activity impressions can be expected to activate each other and elevate recognition performance, whereas dissimilar representations do not interact in a facilitatory way. These results indicate that, depending upon the factors underlying affective impressions, recognition performance can be modified by combinations of impressions. Many researchers discuss the memorability of congruent or incongruent information focusing on schema or expectancy (e.g., Hastie & Kumar, 1979; Brewer & Treyens, 1981). However, little attention has been paid to the role of affective impressions in relation to recognizing pair stimuli with congruent or incongruent impressions. As described above, several studies revealed fundamental differences between characteristics of Evaluation and the other factors of Activity and Potency (Osgood, et al., 1957; Oyama, et al., 1998; Oyama, 2003). Our study also suggests that it is important to focus on the different information-processing bases of the factors, namely sensory or synesthetic processing for Activity or Potency, and subjective or self-relevant processing for Evaluation, which may potentially produce negativity, unexpectedness, or elaboration. Our results will shed light on understanding the recognition of congruent or incongruent information in terms of the affective factors. Further investigation is needed to ascertain exactly how such specific characteristics produce different influences on recognition performance. However, we believe these findings contribute to the comprehension of the processes by which humans recognize or memorize multiple stimuli containing various affective impressions. REFERENCES Bellezza, F. S., & Bower, G. H. (1981). Person stereotypes and memory for people. Journal of Personality & Social Psychology. 41(5), 856–865. Bower, G. H. (1981). Mood and memory. American Psychologist, 36, 129–148. Cacioppo, J. T., & Berntson, G. G. (1994). Relationship between attitudes and evaluative space: A critical review, with emphasis on the separability of positive and negative substrates. Psychological Bulletin, 115, 401–423. Brewer, W.F., & Treyens, J.C. (1981). Role of schemata in memory for places. Cognitive Psychology, 13, 207-230. Chijiiwa, H. (2001). Shikisaigaku gaisetsu. Tokyo, Japan: Tokyo University Press. Cohen, C. E. (1977). Cognitive basis of stereotyping: An information processing approach to social perception. Dissertation Abstracts International, 38(1-B), 412–413. Cohen, C. E. (1981). Person categories and social perception: Testing some boundaries of the processing effects of prior knowledge. Journal of Personality and Social Psychology, 40, 441–452. Collins, A. M., & Loftus, E. F. A. (1975). Spreading-activation theory of semantic processing. Psychological Review, 82, 407–428. Hamilton, D. L., & Gifford, R. K. (1976). Illusory correlation in interpersonal perception: A cognitive basis of stereotypic judgments. Journal of Experimental Social Psychology, 12, 392–407. Hamilton, D. L., & Zanna, L. J. (1972). Differential weighting of favorable and unfavorable attributes in impressions of personality. Journal of Experimental Research in Personality,
6, 204–212. Hastie, R., & Kumar, P. A. (1979). Person memory: Personality traits as organizing principle in memory for behavior. Journal of Personality and Social Psychology, 37, 25–38. Hogg, J. (1969). The prediction of semantic differential ratings of color combinations. The Journal of General Psychology, 80, 141–152. Inoue, M., & Kobayashi, T. (1985). Nihon niokeru SD hou no kenkyubun’ya to sono keiyoushitsui shakudokousei no gaikan [The research domain and scale construction of adjective-pairs in a semantic differential method in Japan]. Japanese Journal of Educational Psychology, 33, 253–260. Isen, A. M., Shalker, T., Clark, M., & Karp, L. (1978). Affect, accessibility of material in memory and behavior: A cognitive loop? Journal of Personality and Social Psychology, 36, 1–12. James, L. A. (1969). The affect of symbols: Towards the development of a cross-cultural semantic differential. International Journal of Symbology, 1, 28–52. Judd, C. M., & Kulik, J. A. (1980). Schematic effects of social attitudes on information processing and recall. Journal of Personality & Social Psychology. 38(4), 569–578. Kanouse, D. E., & Hanson, L. R., Jr. (1972). Negativity in evaluations. In E. E. Jones, D. E., Kanouse, H. H. Kelley, R. E. Nisbett, S. Valins, & B. Weiner (Eds.) Attribution: Perceiving the case of behavior (pp.121–135). Morristown, NJ: General Learning Press. Nakano, M. (1972). Shikisai kanjo to keitai kanjo no gouseikouka ni kansuru bunsekiteki kenkyu [An analytical study of combined affective value of color and form]. Japanese Journal of Psychology, 43(1), 22–30. O'Brien, E. J., & Myers, J. L. (1985). When comprehension difficulty improves memory for text. Journal of Experimental Psychology: Learning, Memory, & Cognition. 11(1), 12-21. Olson, J. M., & James, L. M. (2002). Asymmetrical impact: Vigilance for differences and self-relevant stimuli. European Journal of Social Psychology, 32, 383–393. Osgood, C. E. (1960). The cross-cultural generality of visual-verbal synesthetic tendencies. Behavioral Science, 5, 146–169. Osgood, C. E., Suci, G., & Tannenbaum, P. H. (1957). The measurement of meaning. Urbana: University of Illinois Press. Oyama, T. (2003). Affective and symbolic meanings of color and form: Experimental psychological approaches. Empirical Studies of the Arts, 21(2), 137-142. Oyama, T., Yamada, H., & Iwasawa, H. (1998). Synesthetic tendencies as the basis of sensory symbolism: A review of a series of experiments by means of semantic differential. Psychologia, 41, 203–215. Pratto, F., & John, O. P. (1991). Automatic vigilance: The attention-grabbing power of negative information. Journal of Personality and Social Psychology, 61, 380–391. Sakuta, Y., & Gyoba, J. (2001). Meishi to kao no kanseihyoka to kioku meishi kara kao wo soukisuru baai [Impression judgment and memorability of name cards and faces—Analysis of remembering faces from name cards]. Technical Report of IEICE, 101(422), 7–12. Sakuta, Y., & Gyoba, J. (2003). Kao to meishi no insyo to sorera no kioku youisei [Impression and memorability of faces and name cards]. The Japanese Journal of Mental Imagery, 1(1), 73–80. Smith, D. A., & Graesser, A. C., (1981). Memory for actions in scripted activities as a function of typicality, retention interval, and retrieval task. Memory & Cognition, 9(6), 550–559. Srull, T. K. (1981). Person memory: Some tests of associative storage and retrieval models. Journal of Experimental Psychology: Human Learning & Memory, 7(6), 440–463. Taylor, S. E. (1991). Asymmetrical effects of positive and negative events: The mobilization-minimization hypothesis. Psychological Bulletin, 110(1), 67–85. Taylor, S. E., & Brown, J. D.. (1988). Illusion and well-being: A social psychological perspective
on mental health. Psychological Bulletin, 103, 193–210. Thurstone, L. L. (1931). The measurement of attitudes. Journal of Abnormal Psychology, 26, 249–269. Turvey, M. T., & Fertig, J. (1970). Polarity on the semantic differential and release from proactive interference in short-term memory. Journal of Verbal Learning and Verbal Behavior, 9(4), 439–443. Wickens, D. D. (1970). Encoding categories of words: An empirical approach to meaning. Psychological Review, 77, 1–15. Wickens, D. D., & Clark, S. (1968). Osgood dimensions as an encoding class in short term memory. Journal of Experimental Psychology, 78(4), 580–584. Winograd, E. (1966). Recognition memory and recall as a function of degree of polarization on the semantic differential. Journal of Verbal Learning and Verbal Behavior, 5(6), 566–571. Whittlesea, B. W. A., & Williams, L. D. (2001). The discrepancy-attribution hypothesis: The heuristic basis of feelings of familiarity. Journal of Experimental Psychology: Learning, Memory, and Cognition, 27(1), 3–13. Xiong, M. J., Franks, J. J., & Logan, G. D. (2003). Repetition priming mediated by task similarity in semantic classification. Memory & Cognition, 31(7), 1009–1020. Manuscript received July 14, 2004 Revisions accepted June 20, 2005
APPENDIX
Targets in the recognition test Color 5PB4/10
Form 1
1k
2
2k
3
4
9
9k
10
11
11k
12k
17
19
22
31
34
36
50
52
53
54
57
59
2.5BG8/4 5RP3/4 5P8/4 5PB6/6 2.5RP6/12 7.5BG7/2 5B4/8 2.5G3/4 10YR6/10 7.5PB5/2 2.5R9/2 5YR8/6 7.5R7/8 5YR7/14 2.5YR7/2 10YR7/12 2.5Y8/14 2.5Y9/4 10R3/4 5G7/6 10RP8/4 7.5Y4/4 10B3/6
